MicronTracker

Developer’s Manual
MTC 3.8
October 3 2016

WARNING: This document and the associated demonstration application source code, in
electronic or printed form, are proprietary and confidential to ClaroNav Inc. No part of this
document, the detailed programming interface documentation and the associated example
source code, shall be copied, published or used for a purpose other than product usage
without the prior written consent of ClaroNav Inc.

Part Number 990-MT-USR-GD, Rev E

 MicronTracker Developer’s Manual, MTC3.8 2

Table of Contents
1 Warnings and precautions ............................................................................... 5
1.1 Warnings and precautions ..........................................................................................5
1.2 Safety warnings..........................................................................................................5
1.3 Care ............................................................................................................................6
1.4 Disposal .....................................................................................................................6
1.5 Symbols on the MicronTracker..................................................................................7
2 Product Description and Intended Use........................................................... 8
2.1 Intended Use Statement .............................................................................................8
2.2 Purpose of this Manual ..............................................................................................8
2.3 Third-Generation Optical Tracking ...........................................................................9
2.4 Principles of Operation ............................................................................................11
2.5 Terminology and Conventions .................................................................................12
2.6 Contact Us................................................................................................................14
3 Windows Installation...................................................................................... 15
3.1 MicronTracker parts list ...........................................................................................15
3.2 PC Host Configuration .............................................................................................15
3.3 Enabling the IEEE 1394 Link ..................................................................................16
3.4 Removing an older installation ................................................................................18
3.5 Internet downloadable upgrades ..............................................................................18
3.6 Installing the MicronTracker Camera Device Drivers and USB .............................18
3.7 Installation Verification ...........................................................................................26
4 Linux Installation ........................................................................................... 27
4.1 PC Host Configuration .............................................................................................27
4.2 Linux Installation .....................................................................................................27
5 Mac OSX Installation ..................................................................................... 29
5.1 Release Note ............................................................................................................29
5.2 PC Host Configuration .............................................................................................29
5.3 Mac OSX Installation ..............................................................................................29
6 Product Overview ........................................................................................... 31
6.1 Camera Models and Specifications ..........................................................................31
6.2 Field of Measurement (FOM) ..................................................................................33
6.3 Extended FOM and small marker detection ............................................................34
6.4 Multi-Camera ...........................................................................................................34

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 3

6.5 General-Purpose digital Input/Output (GPIO) .........................................................37

7 Accuracy Performance ................................................................................... 38
7.1 Reporting Terms ......................................................................................................38
7.2 Application-independent accuracy reporting ...........................................................39
7.3 Factors affecting accuracy .......................................................................................39
7.4 Calibration................................................................................................................39
7.5 Drift Factors .............................................................................................................41
7.6 Jitter .........................................................................................................................44
8 The VB.Net Demo Application ...................................................................... 45
8.1 Camera Selection .....................................................................................................46
8.2 The Activate Menu ..................................................................................................46
8.3 The Display Menu....................................................................................................47
8.4 The Video Recorder Menu.......................................................................................48
8.5 Help ..........................................................................................................................49
8.6 The Marker Templates Dialog .................................................................................49
8.7 The Pose Recorder ...................................................................................................59
8.8 The Snapshots Dialog ..............................................................................................60
8.9 The 3D Tracer Dialog ..............................................................................................61
8.10 The Options Dialog ..................................................................................................62
9 The R-Fine Application .................................................................................. 64
9.1 R-Fine Overview ......................................................................................................64
9.2 Main screen ..............................................................................................................65
9.3 The Application menu..............................................................................................66
9.4 The Display menu ....................................................................................................66
9.5 The Help menu.........................................................................................................67
9.6 The Camera ..............................................................................................................67
9.7 The Results window.................................................................................................67
9.8 The log file ...............................................................................................................69
9.9 R-Fine Quick Guide .................................................................................................69
10 Marker Design ................................................................................................ 72
10.1 Introduction ..............................................................................................................72
10.2 XP contrast ...............................................................................................................72
10.3 XP flatness ...............................................................................................................73
10.4 Xlines .......................................................................................................................73
10.5 Facet Design.............................................................................................................74

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 4

10.6 Multi-facet Marker ...................................................................................................77

10.7 Reference marker design..........................................................................................79
10.8 Design examples ......................................................................................................79
10.9 Materials ..................................................................................................................81
10.10 To learn more ...........................................................................................................82
11 Programming Interface.................................................................................. 83
11.1 Programming Model ................................................................................................83
11.2 Detailed C API Description .....................................................................................85
11.3 Source Code Examples ............................................................................................85
11.4 The SimpleDemoC Application ...............................................................................86
11.5 The MTDemoCPP Application................................................................................87
12 Advanced Topics............................................................................................. 88
12.1 Camera Exposure Controls ......................................................................................88
12.2 Overexposure Control Interleave .............................................................................91
12.3 Templates Matching Tolerance ................................................................................91
12.4 Predictive Frames Interleave ....................................................................................92
12.5 Reducing CPU Load ................................................................................................92
13 Standards......................................................................................................... 94

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 5

1 Warnings and precautions

1.1 Warnings and precautions
NOTE: Please read this manual and follow its instructions carefully. The manual
assumes that the MicronTracker developer is familiar with pose tracking systems
in general and has sufficient background of spatial kinematics and mathematics.

MicronTracker’s measurement accuracy could be compromised under

circumstances such as: operating temperature out of calibrated range; insufficient
lighting conditions; camera not fully warmed-up; measurement outside calibrated
field; and Shadow over XPoints. When conditions which may affect the
measurement accuracy are detected, a warning message is provided by the
MicronTracker software. Please see Microntracker API documentation for the
complete list of warnings returned by MT functions call.

MicronTrackers’ interface library, MTC, provides no user interface. Handling of

measurement warnings and runtime program exceptions or errors is the
responsibility of the application product development team.

1.2 Safety warnings

WARNING: No Patient Applied Parts

MicronTracker is a low voltage electrical device, which is not to be put in contact
with the patient (No Patient Applied Parts). MicronTracker has been tested to
comply with FCC Part 15, Class A specification and to be conformant with EN
60601-1-2: General Requirements for Safety—2. Collateral Standard:
Electromagnetic Compatibility—Requirements and Tests.

WARNING: Check enclosure electrical current leakage

in end use application
Since the MicronTracker can be used with various types of power supplies,
camera enclosure electrical current leakage is to be measured in the end use
application.

WARNING: Do not remove silicon case

MicronTracker camera is equipped with a silicon case to prevent camera from
ESD discharge. ESD discharge could freeze image streaming or corrupt EEPROM
on the camera. Make sure the GPIO port is covered by the GPIO cap when it is
not used. If you are mounting camera on a tripod, make sure the tripod screw
wouldn’t transmit ESD to the camera.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 6

WARNING: Do not open camera case

MicronTracker camera is optically pre-calibrated at the factory and should not be
disassembled or modified in any way. Opening, or attempting to open, the camera
case will compromise the measurement performance of the device and invalidate
its warranty. An optional recalibration service is available from ClaroNav Inc.by
contacting support at the address in Section 2.6.

WARNING: Protecting electronics of the camera

Extended exposure to bright sunlight, rain, dusty environments, etc. may cause
problems with the electronics and the optics of the system. Static discharge could
damage the camera. The camera silicon case is designed to protect the camera
against static discharge and should not be removed. Connecting firewire cable in
the wrong direction could damage the physical layer of the camera. Using a
FireWire card other than the one shipped with the camera or from Point Grey can
cause unexpected behavior and possible damage to the camera.

1.3 Care

Handling
MicronTracker is a precise optical instrument and should be treated as such. When
handling the camera unit, avoid touching the lenses. Avoid excessive shaking,
dropping or mishandling of the device. The case is designed for maximum
protection, but when not in use keep the MicronTracker in its foam-padded box
and avoid shocks.

Cleaning
Keep the lenses free of dust or stains using a photographic lens cleaning solution
and a lint free paper or cloth. Do not apply excessive force.

1.4 Disposal

MicronTracker should not be treated as household waste. It may be

returned to our facility for proper disposal. Contact Technical Support to return a
product for disposal by obtaining return authorization. There is no charge for the
recycling of the returned products; however the cost incurred for returning
products to ClaroNav is the responsibility of the customer.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 7

1.5 Symbols on the MicronTracker

Read the Developer’s Manual

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 8

2 Product Description and

Intended Use
2.1 Intended Use Statement
MicronTracker products provide fully passive real-time sub-millimeter optical
measurement of the pose (location and orientation) of specially marked objects
within their measurement field. The platform is used in products for the tracking,
placement and operation of instruments, tools or device positional planning and
operation.

MicronTracker products are verified through standard testing methods with

product risks having been evaluated on a general level only. Verification,
validation and / or product risks evaluation specific to any clinical use case or
within the context of any given intended use scenario shall be performed by the
developers of MicronTracker-based application products, and not by ClaroNav
Inc.’s MicronTracker development team.

It is the full responsibility of the application product developer to correctly use the
measurement data provided a MicronTracker product, and for validating the
suitability and safety of the system incorporating a MicronTracker in the overall
system’s intended use.

2.1.1 Product Description

MicronTracker is a family of real-time (at least 10Hz) sub-millimeter optical pose
tracking products. The products use visible light to detect and track object of
interest. These objects are marked using small checkered target regions called
XPoints. The MicronTracker camera is connected to a PC via a standard interface
and image data is transferred via this interface and processed on the PC to obtain
the measurements reported to a client software application.

The product is aimed to be integrated into a complete application by developers,

and is not intended for use directly by clinical end users. Application developers
should select the right MicronTracker model based on considerations such the size
and distance of the field of measurement, minimum measurement rate needed, and
accuracy required and price. The specifications for each model are easily
accessible and should be consulted during model selection.

2.2 Purpose of this Manual

The purpose of this manual is to assist you in effectively using your new
MicronTracker product. In particular, you will learn how to:
 Install the MicronTracker software on a workstation or a laptop.
 Design markers optimized for use with the MicronTracker.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 9

 Plan the integration of MicronTracker’s software into your application.

 Optimize the performance of the MicronTracker to better fit the requirements
of your application.
A detailed description of the programming interface is separately provided in the
MTC programming manual, in HTML format.

2.3 Third-Generation Optical Tracking

MicronTracker is a family of optical pose tracking products, all sharing a common
software interface, the MTC library, but differing in the properties of their camera.
The two key functions of optical pose trackers are:
 Detection: detect the presence of, i.e., identify, specially marked objects in
the sensor’s field of measurement (FOM).
 Pose measurement: report on the location and orientation of each detected
object.
Both functions need to be performed repeatedly at real-time rates (10Hz or faster).
Optical pose trackers perform these functions by observing marked locations, or
targets, on each tracked object from multiple angles of view, then triangulating
the lines of sight to the targets to calculate each target’s’ location. The 3D
locations of at least 3 targets are needed to calculate the object’s pose, i.e., its
position and orientation in space relative to the camera. Pose sensing is sometimes
referred to as 6D measurement, since at least 6 parameters are required to fully
describe a pose: 3 space coordinates (X, Y, Z) and 3 orientation angles (e.g.,
azimuth, elevation and roll).
MicronTracker is the first third generation1 optical pose tracking product (see
illustration). Optical pose sensors of the first generation use light-emitting targets,
such as LEDs (light emitting diodes). Second-generation sensors emit light from a
ring surrounding each camera lens and use balls or discs as targets. The balls or
discs are coated with a retro-reflective material that mirrors the light back to the
lenses. Both first and second generation trackers require the targets to be observed
as bright spots over a dark background. Since this cannot be guaranteed under
visible light conditions, Infra-Red (IR) light is used. The intensity of the light
reaching the camera falls with the square of the distance between the target and
the camera. To prevent sensor saturation, first and second generation trackers do
not allow targets to be placed close to the camera. Placing the camera far from the
FOM means that a larger camera is needed to obtain the required accuracy. It also
increases the likelihood of line-of-sight interruptions and makes it more difficult
to optimally position the camera, especially when space is at a premium.

1
ClaroNav Inc. holds patents on key enabling technological innovations related to 3 rd
generation optical tracking.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 10

Third generation trackers are fully passive. They use available visible light
illumination to observe targets stereoscopically much like humans do. Tracked

1st Generation: active IR targets

IR camera I
R
Application computer
C
IREDs
a
m
e
r
a
Custom electronics

IR camera +
IR light
2nd Generation: active IR camera

Retro-reflective
balls

Custom electronics

3rd Generation: fully passive

Camera

Visible pattern

objects are marked with a visible painted or printed target pattern. The camera is
interfaced to the computer using a standard digital link (e.g., FireWire or USB
2.0). Software running on the PC processes the camera images to detect and
pinpoint visible target patterns in the images. The MicronTracker’s target pattern
is a set of high-contrast regions called a Marker. The regions forming the marker
are geometrically defined and arranged according to specific rules dictated by the
tracker’s designers.
Third generation trackers offer many advantages over trackers of earlier
generations, amongst them:
 Simpler, more reliable, lower-cost, design, which uses off-the-shelf
electronics;
 Smaller and lighter camera that can be placed closer to the FOM, easier to
ship, move, store and manipulate;

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 11

 Lower cost markers, which can be easily produced or durably painted on

objects;
 Better immunity against measurement errors resulting from partial occlusion
of targets;
 More reliable identification of marker patterns;
 Practically unlimited number of objects tracked concurrently;
 New target patterns can be easily learned by the software in the field;
 Great flexibility in optimizing the designs of the cameras and target patterns
to fit specific application requirements;
 Easy product upgrades in the field through simple downloadable software
updates;
 Application software access to camera images for easier set up, augmented
reality display, documentation and remote diagnostics;
 Support for multi-camera configurations, used to eliminate line-of-sight
interruptions or expand the FOM.
Over the next few years, the third generation trackers will undergo rapid
evolution. In particular, expect to see markers becoming smaller and less
intrusive, accuracy and measurement rates improve, and many additional software
features added to allow configuring and customizing trackers to the specific needs
of users in a wide range of applications.

2.4 Principles of Operation

Detection of markers is achieved by processing
the images of the two sensors, left and right, to
detect and correlate visual patterns matching the
descriptors in the marker templates database.
Each target point in a marker has to be correctly
recognized in each of the two images. This
requires that the checkerboard intersection
region cast a minimum projection “footprint” in
each image. When the smallest target region
projection width falls below the minimum
required footprint diameter (currently
configurable to 11 or 9 pixels), the target is not
detected. Camera models with higher sensor
resolution or longer lenses (narrower field of
view) generate larger projection image footprints, and are, therefore, able to
recognized smaller targets, or targets at larger distances. Tilting a target region
away from facing the camera reduces the width of its footprint proportionally to
the tilt angle’s cosine.
Once a marker pattern has been identified in the images, the exact 3D position of
its target points is calculated by triangulating the two projection lines associated
with the two image positions in which the target center is observed (see diagram).
These projection lines are obtained from a calibration file, computed from data
collected in a series of experiments carried out before each camera is shipped
from the factory.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 12

The pose of each marker is obtained from

measuring the individual 3D positions of its Vector
target points, then calculating the coordinate
B vector
space transformation that best matches the
template coordinates of each target to the W
camera-space coordinates measured, i.e.,
resulting in the minimum RMS error between
measured and transformed locations.

2.5 Terminology and Conventions

Unfortunately, when it comes to optical tracking, there is no widely-accepted
terminology. This is especially true when it comes to a new generation of
products. To reduce confusion, this section defines the terms being used in the
MicronTracker development environment.

2.5.1 Data Capture

(Stereo) Camera: The stereo camera assembly.
Lens/sensor: The optical/electronic part of the camera involved in producing
images.
Left/Right Image: The pixels buffer acquired from each corresponding
lens/sensor pair.
(Stereo) Frame: A pair of Left and Right images captured simultaneously
by the same camera.

2.5.2 Object marking

Target: a single physical location whose position in 3-space is being marked
and measured. This is a general term used in
optical metrology, usually referring to the
center of retro-reflective discs or balls. WB Xline
MicronTracker products use Xpoints to mark
targets (see next). Xpoint

Xpoint (or XP): one checkerboard

intersection painted on a flat surface. The term
is also used to refer to the appearance of such a
BW Xline
pattern in an image.
Xlines: The two imaginary straight lines
crossing at the Xpoint's center. Viewing the
Xlines from the Xpoint center outwards, BW
Xline is the one with the black on the left and white on the right (hence the letter
order), and WB Xline is the other. The Xlines are usually designed to be
perpendicular to each other, but this is not a requirement.
Xnormal: A unit vector perpendicular to both Xlines, on the painted side of the
Xpoint’s surface (see the multi-facet marker figure on next page).
Vector: Two Xpoints arranged such that one's BW Xline is aligned (co-linear)
with the other's WB Xline. The Xpoint at the BW end is called the vector's base,

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 13

and the other is the vector's head. Although not an absolute restriction, the
Xnormals of the XPs at the ends of a vector would usually be identical (or at least
very similar). Vectors in the FOM are detected and tracked, even when they are
not identified as being part of a facet.
Facet: Two vectors in a rigid configuration that
complies with MicronTracker’s facet restrictions (one Marker
vector is longer than the other, and the angle between
the vectors is in the 8-172 degrees range, see 10.5). The 1..N
vectors would usually be on the same plane, or at least F
have parallel Xnormals, but this is not a requirement. A F
Facet
a
facet, containing 3 or 4 Xpoints, is the smallest set of a
shorter cc
longer
Xpoints that can be tracked in 6D (3 spatial coordinates e
and 3 axes of rotation). e
t
Vector tV
Marker: One or more facets in a rigid configuration, e
usually attached to an object to be tracked. In a multi- base headc
facet marker, only one of its facets needs to be t
Xpoint X
o
observed in a frame to allow MicronTracker’s software
P
r
to detect and report on the marker’s pose.
o
Marker Template: A named description of a marker’s geometry, enabling i
its detection and pose measurement. Marker templates are stored on disk as n text
t

A Multi-facet marker attached

Facet A (2 red vectors) to a pointer tool
x y

z Xnormal

Facet B (2 green vectors)

z y
x

files.

2.5.3 Coordinate systems

All coordinate systems are right-handed. Locations and distances are reported in
millimeter units.
Facet coordinates: Each facet has its own coordinate system. The origin is
at the middle of the longer vector. The X-axis extends from the origin towards the
head of the longer vector. The Z-axis is perpendicular to both vectors, pointing in
the direction opposite to the facet’s Xnormals, such that it is aligned with the

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 14

camera’s Z-axis (depth direction) when the facet is facing the camera. The Y-axis
is defined according to the right-handed coordinate system’s rules (see above
figure for examples).
Marker coordinates: Identical to the coordinates of the first facet described
in the marker’s template (even when that facet is not visible).
Camera coordinates: Defined during calibration. The origin is
approximately at the center of the camera. The Z-axis extends to the front of the
camera (depth), X towards the right of the camera and Y down. When viewing the
camera’s images, the camera and the screen coordinate systems are aligned (X
right, Y down and Z in).

z
y

NOTE: The camera coordinate axes orientations are not accurately aligned with
the orientation of the enclosure’s external faces, and the FOM is not accurately
symmetric around each axis.

2.6 Contact Us
It is quite possible that the environment in which you will be using your
MicronTracker is different from any of the environments in which it was tested in
the past, and, thus, you may encounter some difficulties or have new requirements
that have not been fully addressed yet. We, at ClaroNav Inc., are keenly interested
in your experience evaluating the device and embedding it in your application. We
commit to being highly responsive to any feedback we receive.
Our contact coordinates:

ClaroNav Inc.,
1140 Sheppard Avenue West – Unit 10
Toronto, Ontario, Canada, M3K 2A2
Tel: +1.647.951.1525, Fax +1.647.951.1524
www.claronav.com
[email protected]

Authorized Representative (for Regulatory Affairs in Europe):

Emergo Europe
Molenstraat 15
2513- BH The Hague
The Netherlands
Tel: (31) (0) 70 3458570

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 15

3 Windows Installation
3.1 MicronTracker parts list
The MicronTracker package contains:
 One MicronTracker camera
 One IEEE 1394/FireWire PCI Host Card (Supports serial bus data transfer
rate of 100/200/400Mbps and 800Mbps for H3-60)
 One 4.5m 6 pins to 6 pins IEEE 1394 (FireWire) cable only for Sx60, Hx40
and Hx60 models
 One 4.5m 9 pins to 9 pins FireWire 800 IEEE 1394b cable only for H3-60
model.
 One General Purpose Input Output (GPIO) connector (Only Hx40, Hx60,
Sx60 and H3-60 models)
 The MicronTracker USB flash drive
 A board demonstrating two markers, one of them supporting tooltip
calibration
 A polished aluminum bar (AccBar) to be used with R-Fine software for
verification and refinement of the calibration
 A special CoolCard marker, made of the Duramark material, for adjusting the
light coolness camera setting
 Printed sheets showing examples of marker patterns
 Mini tripod
If your package does not contain the any of the above items, please contact
ClaroNav Inc.support at: [email protected]

3.2 PC Host Configuration

To be able to install and properly run the MicronTracker software, the following
system configuration is required:
 MS-Windows 7/Win 8/Win 10, Linux, 32bit and 64bit
 2GHz or faster Pentium processor (3GHz recommended)
 1GB RAM
 1GB hard disk space
 An available IEEE 1394 (FireWire) socket or an available PCI slot for the
IEEE 1394 card provided with the MicronTracker.
To be able to run and modify the source code of the demonstration programs:
 Display of 1024x768 or higher resolution
 1 GB hard disk space and 1GB of memory
 C++ or Microsoft .Net development environment

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 16

3.3 Enabling the IEEE 1394 Link

3.3.1 IEEE 1394 (FireWire) connectors
The MicronTracker camera communicates with the PC host using an IEEE 1394
(also known as FireWire or i.link) digital link. IEEE 1394 is a fast serial link
commonly used to connect consumer digital video cameras to computers. There
are three types of IEEE 1394 connectors: a wide 6-pin, narrow 4-pin and IEEE
1394b 9-pin. The three types of connectors and their matching sockets are shown
below.

MicronTracker is shipped with a 6-pin to 6-pin or 9-pin to 9-pin cable (depending

on the model) and an IEEE 1394a PCI board (which has two 6-pin IEEE 1394
ports and one 4-pin IEEE 1394 port) or IEEE 1394b PCI board (which has two 9-
pin IEEE 1394b ports and one 6-pin IEEE 1394a port) to use with stationary PCs
that do not already have a proper socket available.

3.3.2 Connecting the camera to an unpowered socket

The MicronTracker camera obtains its power through the IEEE 1394 cable. The 4-
pin socket, which is commonly found on laptops, and some 6-pin sockets on
mobile computers, does not provide power. A powered IEEE 1394 hub or repeater
is, therefore, needed between the camera and the laptop.
Alternatively, it is possible to inject power directly into an unpowered IEEE 1394
connection using a special cable with a power jack. The example shown below is
RR-CFR-64P-01 by Newnex, available from usbfirewire.com.

NOTE: Ensure that you use a power adapter with enough power to enable the
MicronTracker to function properly (12VDC, 2.5W per camera).

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 17

3.3.3 Installing the IEEE 1394 PCI Board

Figure 1: The Device Manager dialog, showing that the computer detected the 1394 bus
controller and the PCI card is correctly installed.
To install the IEEE 1394 OHCI PCI host adapter 3-port 400 MB/sec or 2-port
1394b 800Mbps you will need complete the following steps:
1. Make sure that the computer is turned off.
2. Unplug the power cord from the computer.
3. Remove the side panel from the computer.
4. Place the card in an open PCI slot. Making sure that it is firmly seated. Note
that using 1394b card some of the pins will remain outside of the PCI slot (see
below).

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 18

5. It is also recommended to connect the power pin on the Firewire card to the
power supply on the mother board.
6. Replace the side panel on the computer.
7. Plug back the power cord and turn the computer on.
8. The screen will show a New Hardware Found dialog box. Then an Add New
Hardware Wizard box will appear. Follow the prompts given by the Wizard.
9. To ensure that the PCI card is properly installed, open the Windows Control
Panel, click on the System icon, then the Hardware tab and finally select
Device Manager. The presence of the IEEE 1394 Bus host controller entry as
shown in fig. 1 indicates that the PCI card has successfully been installed.

3.4 Removing an older installation

When upgrading an existing MicronTracker installation with a newer version, first
it is recommended to uninstall existing cameras. With the camera connected, open
the Device Manager and look under Imaging Devices for a device named “PGR
Bumblebee”. Right click on the device and select uninstall. After uninstalling the
device, unplug the camera. Proceed to uninstall the previous MicronTracker
software version by navigating to Control Panel  Add or Remove Programs 
MicronTracker.
Since the device driver may have been loaded in memory, it is recommended to
reboot the machine before starting to use the new MicronTracker version.
NOTE: On windows platforms, the camera driver is installed under
Windows/system32/drivers folder and uninstalling the software doesn’t remove
the driver file. For a clean uninstallation, delete “PGR1394.sys” and pgrcam.sys
from system32\drivers folder.

3.5 Internet downloadable upgrades

When upgrading an existing MicronTracker installation (version 2.0 or higher)
with an installer downloaded from the Internet, it is not required to uninstall the
previous version. Simply follows the steps of the installer in sequence, as detailed
below. When installing the MicronTracker package, select the “Repair” option
when asked. This different procedure is necessary since the camera specific
calibration file is not included in the downloadable upgrade, and therefore the
existing installation needs to be already in place. Alternatively the camera
calibration file (BumbleBee_xxxxxx.calib or FlyCapture_xxxxxx.calib, where
xxxxxx represent the serial number) can be copied in the folder
[INSTALLER_ROOT]\ClaroNav\MicronTracker\CalibrationFiles to transform the
update installer in a complete camera specific installer.

3.6 Installing the MicronTracker Camera Device

Drivers and USB
The MicronTracker Installer available on the MicronTracker USB will install the
necessary camera drivers, a set of complementary libraries from Point Grey
Research, the MicronTracker interface software and several demonstration
applications and their source code.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 19

NOTE: Since the procedure involves installing device drivers, the user performing
the installation needs to have administrative privileges. Otherwise, the installation
may fail.
NOTE: Do not connect the camera to your computer until you have completed the
software installation steps.

3.6.1 Installer Steps

The installer provided on the MicronTracker USB includes some basic installation
instructions, and references this document (available on the USB before the
installation is performed) for any further information.
The MicronTracker distribution contains FlyCapture SDK, library developed by
Point Grey Research (see www.ptgrey.com for details). Its installation is required
for MicronTracker to function properly.
As part of the pre-requisite installations, the FlyCapture SDK will be installed
prior to the MicronTracker installation (fig. 2).

Figure 2: The Micron Tracker installation interface. You will be prompted to install
FlyCapture before the Micron Tracker installation proceeds.

The installer provides 32bit or 64bit MicronTracker releases for 32bit or 64bit
operating systems. The appropriate FlyCapture and MicronTracker will be
installed on your system.
FlyCapture SDK is the low level video libraries and it also installs all the required
drivers.
Simply click on the appropriate menus on the main installer and then follow the
instructions of the individual installers.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 20

NOTE: The installer will detect whether your Windows is 32 or 64 bit and install
the appropriate version. Those who wish to develop 32 bit applications on their 64
bit machines will still be able to do so, as the appropriate files are still installed.

3.6.2 Installing MicronTracker

Click on the menu “Install MicronTracker”.

Figure 3: The installShield Wizard for MicronTracker

After the Welcome screen a dialog to select where the application folder should be
located is presented to the user.

Figure 4: Select the MicronTracker’s installation directory

Then a dialog to select the program folder will be presented.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 21

Figure 5: Select the MicronTracker’s program folder

After that, the installer will start transferring files. When terminated, the installer
will display the Wizard Complete dialog (fig. 6).

Figure 6: MicronTracker has been installed successfully

3.6.3 Installing FlyCapture

When prompted to install FlyCapture during the MicronTracker installation, select
“Yes”. The FlyCapture installer will appear and select “Complete” installation
type. This option will install drivers, dlls, documentation and sample codes.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 22

Figure 7: Select the “Complete” installation type

Then select the entire features to be installed.

Figure 8: Select the entire features to be installed

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 23

MicronTracker Camera is a firewire camera, select “firewire camera” drivers to

install.

Figure 9: Select the “firewire cameras” driver to install

Then confirm the cameras driver selection.

Figure 10: Confirm the camera drivers to install

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 24

To optimize perfromance, let flycapture to control processor idle state. This is

recommended to avoid any image tearing issue.

Figure 11: Disable processor idle state when using FlyCapture

Once Installation is complete, it is recommended to reboot the machine.

3.6.4 Connecting the Camera

To connect the camera to the computer, perform the following steps:
1. Plug one end of the cable into one of the sockets on the newly installed PCI
card or IEEE 1394 hub (it does not matter which port you plug it into).
2. Plug the other end of the cable into the socket on the back of the MicronTracker
camera.
When the camera is connected and powered, the green LED at the front of the
MicronTracker camera will glow. You may need to wait a minute or two before
being prompted to install the camera driver.

3.6.5 Installing the Camera Driver

When you plug the MicronTracker camera into an available IEEE-1394 port for
the first time, you might be presented with the following dialogs to install the
camera driver.
When presented with the Welcome to the Found New Hardware Wizard dialog
click Next.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 25

Figure 12: Found New Hardware Dialog box, click Next to continue

In Windows 7, you will then be presented with a Warning dialog informing you
that the camera driver is not Microsoft certified. Click “Continue Anyway” and
ignore the warning. (fig. 8).

Figure 13: Microsoft uncertified driver warning, just ignore it and click Continue Anyway
Complete the process by clicking Finish on the Completing Found New Hardware
Wizard dialog.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 26

Figure 14: The camera driver has been installed successfully

3.7 Installation Verification

After the MicronTracker installation is complete, open the DriverControlGUI
application (it is installed as part of FlyCapture and can be found under
FlyCapture installation directory) and verify that the 1394 devices connected to
the camera is using Point Grey research drivers. The application also allows
switching between 1394 device drivers provided by other vendors as well.

Figure 15: DriverControlGUI, Lists all 1394 devices and their drivers

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 27

4 Linux Installation
4.1 PC Host Configuration
To be able to install and properly run the MicronTracker software, the following
system configuration is required:
 A Linux distribution with kernel 3.2 or higher including gcc. Previous
versions of the kernel had inconsistent FireWire support.
 2 GHz or faster Pentium processor (3GHz recommended)
 500 MB RAM
 200MB of available hard disk space
 An available IEEE 1394 (FireWire) socket or an available PCI slot for the
IEEE 1394 card provided with the MicronTracker.
 Display of 1024x768 or higher resolution

4.2 Linux Installation

The MicronTracker SDK on Linux is distributed as a tgz or zip file in conjunction.
The MicronTracker SDK also contains a multi-platform demo. This demo is not
as complete as the Windows demos, but it shows how to perform the fundamental
MicronTracker tasks:
 Loading pre-existing marker templates,
 Visually generating new templates,
 Recognizing markers,
 Displaying positions and distances between markers.
The UI is has been implemented using FLTK (www.fltk.org) a public domain
multi-platform toolkit. The video and text display is done using OpenGl or FLTK
depending on configuration parameters. Note that the display of images and
overlays is just an example implementation. Cross-platform compatibility was
given the highest priority.

4.2.1 Dependencies
The Linux release comes in the form of a gzip tar archive. After extracting the
sources and libraries to the filesystem you need to:
o Run "Ubuntu Software Center" and install the following development
libraries:
a) libvxl1-dev
b) libfltk1.3-dev
c) libxft-dev
o Set the MTHome environment variable to where you have extracted the
tarball. An example environment.sh file is included for your convenience.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 28

o You will also find an empty CalibrationFiles directory at the higher level
where you extract the tar file. You need to copy the calibration file of your
camera from the USB into this directory. The directory where you’ll find it on
the USB is:
CalibrationFiles
And the file will be named:

BumbleBee_xxxxxx.calib
FlyCapture_xxxxxx.calib
Where xxxxxx will be the serial number of your MicronTracker.

o The release includes a precompiled demo in the "bin" directory. select the
corresponding x64 or x32 bin directory. Make sure that execute permissions are set.
Issue these commands
chmod +x mtsimpledemoc
chmod +x mtdemocpp

o Now the binary files should be executable. issue the following command to run the
simpledemoc app:
./mtsimpledemoc

o The Source code of the MTDemoCPP and MTSimpleDemoC are available in the
NetbeansProjects folder. You would need to install JDK7u10 and Netbeans 7.2.1 for
CPP development in order to open these projects, compile and build them.

Additional useful third party libraries are included in the tarballs folder. They are
not required for using MicronTracker, but you may find them useful when
experimenting with other uses for the stereo video input, such as augmented
reality.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 29

5 Mac OSX Installation

5.1 Release Note
Please note that MicronTracker version 3.7 doesn’t support Mac OSX, however
the package includes the legacy Mac OSX version 3.6. This chapter describes
installation process for MicronTracker version 3.6 on Mac OSX version 10.6.8.

5.2 PC Host Configuration

To be able to install and properly run the MicronTracker software, the following
system configuration is required:
 A Mac OSX 10.6.8 distribution with Xcode and gcc.
 2 GHz or faster Pentium processor
 512 MB or more RAM
 200MB of available hard disk space
 An available IEEE 1394 (FireWire) socket or an available PCI slot for the
IEEE 1394 card provided with the MicronTracker.
 Display of 1024x768 or higher resolution

5.3 Mac OSX Installation

The Mac release comes in the form of a package file and has been tested on Mac
OSX 10.6.8. After extracting the source and libraries you need to:
 Set the MTHome environment variable to where the package has been
extracted. An example environment.plist file is included for your
convenience.
 You will also find an empty CalibrationFiles directory at the higher level
where you extract the package file. You need to copy the calibration file
of your camera from the USB into this directory. The directory where
you’ll find it on the USB is:
CalibrationFiles
and the file will be named:
BumbleBee_xxxxxx.calib
FlyCapture_xxxxxx.calib
where xxxxxx will be the serial number of your MicronTracker.
 The release includes a precompiled demo in the directory MTDemoCPP
(with the Xcode project), but you can invoke make at the higher level to
compile everything from scratch.

NOTE:

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 30

 MTDemoCPP invokes MTHome; an environment variable; to load the

calibrationfile and the Markers file.
 The Xcode sample project has been tested on Xcode version 2.4.1. If you
are using different versions make sure that all of the required
dependencies are assigned properly.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 31

6 Product Overview
6.1 Camera Models and Specifications
MicronTracker cameras are customized versions of PointGrey’s class-leading
Bumblebee stereo cameras. Four models are currently offered, XB3-BW-H3-60,
BB2-BW-Sx60, BB2-BW-Hx40 and BB2-BW-Hx60 (Hx40, Sx60 and Hx60 are
drop-in replacements for the original H40, S60 and H60 models and feature higher
frame rate and GPIO connector for external trigger and strobe functionality). The
following table describes and compares their key characteristics.
Specification \
H3-602 Sx60 Hx40 Hx60
Model
FOM, spherical
section 240x200x160 200x130x100
115x70x55 cm 120x120x90 cm
(radius x width x cm cm
height)
Measurement rate 16 Hz 48 Hz 20Hz 20Hz
Calibration
0.20 mm RMS 0.25 mm RMS 0.20 mm RMS 0.35 mm RMS
accuracy1
Jitter, static target2 0.007mm RMS 0.007mm RMS 0.015mm RMS 0.015mm RMS
Jitter, moving
0.07 mm RMS 0.07 mm RMS 0.14 mm RMS 0.14 mm RMS
target2
Processing
time/frame
15 ms 7 ms 10 ms 10 ms
(2.5GHz CPU
Q8300, 4 markers)
Lag ~60 ms ~20 ms ~50 ms ~50 ms
Sensors resolution 1280x960 640x480 1024x768 1024x768
Lenses / H°xV° 6 mm 50°x38° 6 mm 50°x38° 4 mm 70°x52° 6 mm 50°x38°
Case dimensions 283x43x49 164x43x54 164x43x54 164x43x54
(WxHxD, pprox..) mm mm mm mm
Weight ( pprox..) 505gr 342gr 342gr 342gr
IEEE-1394b
Electrical Interface IEEE-1394a (FireWire), 400 Mbps
(FireWire), 800 Mbps
Voltage
8-30V via IEEE-1394 interface or GPIO connector
requirement
Power consumption 4W at 12V 2.5W at 12V
Mount ¼” thread tripod mount
Operating
18° to 30°C (64° to 86°F)
temperature
Operating 700 - 1060 hPa (up to 2000 m)
pressure/altitude
Operating humidity 30% -80% non-condensing
Storage
-30° to 60°C (-22° to 140°F)
Temperature
Storage humidity 10% - 90% non-condensing

2 Last four characters of the camera’s full name are used as short form.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 32

Warm-up period ~20 mins ~15 mins

Warm-up period
Template-based ~10secs ~10secs
correction enabled
Max. # of marker
practically unlimited
templates
Max. # of markers
concurrently 100
tracked
Illumination range 50-100,000 Lux (20-400,000Lux in HDR mode)
Max illumination
variation between 6:13 (1000:1 in HDR mode)
markers

Certifications IEC 60601-1, FCC Part 15 Class A, RoHS and WEEE compliant

Drift: temperature,
pressure, light
intensity, marker
insignificant
orientation, motion
blur, exposure
settings
Hazard warnings Marker out of FOM; Thermal instability; Temperature out of range; Shadow over
issued target
Operating system MS-Windows7/Win 8/Win 10, Linux, Mac OSX – 32bit and 64bit
Minimal computer
2 GHz CPU, 1GB RAM, 1GB disk space
hardware
NOTES:
1) Single target position accuracy. For Hx40, ~30,000 averaged positions at distances
of 40-85cm, for Sx60, ~10,000 averaged positions at distances of 50-85cm, for
Hx60, ~30,000 averaged positions at distances of 60-155cm, for H3-60, ~50,000
averaged positions at distances of 60-170cm. Calibration accuracy decreases when
smaller footprint detection is enabled.
2) Single target at a distance of 75cm, 150 Lux, 20ms shutter.
3) The illumination variation range can be extended to 20:1 with the use of black/gray
markers (see 10.2).

The camera’s optics and electronics are housed in an internal aluminum case and
an external rubber shell. The external shell is designed to protect the internal case
against electric and physical shock. The camera has been certified for compliance
with medical safety standards and carefully calibrated to perform with the external
shell.
NOTE: Removing the protective shell would nullify the certification and degrade
the camera’s measurement accuracy.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 33

6.2 Field of Measurement (FOM)

MicronTracker’s FOM is defined as a volume in which Xpoint targets of a 12 mm
radius can be detected within accuracy specifications. The FOM boundaries are
shaped approximately as a rectangular section of a sphere, as shown below.
MicronTracker supports both the standard and extended FOM size (shown in blue
and red, respectively, in the below diagrams). The size of the FOM can be
extended as required, by enabling Xpoint detection with smaller image footprint
(details in Section 4.2.1).

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 34

6.3 Extended FOM and small marker detection

An Xpoint target must cast a minimum projection “footprint” in the left and right
camera images, in order to be detected by MicronTracker. MTC API provides
means to configure the minimum footprint size to either 11 or 9 pixels in
diameter. By reducing the minimum detectable projection of targets, the
MicronTracker’s field of measurement can be extended, so that Xpoints can be
detected at larger distances, or when tilted at larger angles away from facing the
camera. Alternatively, the smaller footprint option can be used to enable detection
of Xpoint targets of a smaller physical diameter within the standard FOM. The
tradeoff for such an increased target visibility is slightly lower detection accuracy,
since fewer pixels may be used to pinpoint the position of an Xpoint.
All of the demo applications provide the interface for choosing between the two
footprint sizes. In particular, “Detect small markers” option in the Options dialog
box enables the 9 pixel footprint detection. Note that this option is enabled by
default.

6.4 Multi-Camera
MicronTracker’s software is designed to allow interfacing to a multi-camera set-
up with ease. Starting from MicronTracker version 3.7, each camera requires full
bandwidth to operate properly. Hx40, Hx60 and Sx60 require 400Mbps bandwidth
and H3-60 model requires 800Mbps bandwidth. Firewire 1394a cards provide
400Mbps bandwidth therefore each 1394a camera must be connected to a separate
firewire card. Firewire 1394b cards provide 800Mbps bandwidth; therefore it is
possible to connect two 400Mbps cameras to the same card. This design
guaranties each camera will work at full framerate.
Note that Cameras on separate buses do not automatically synchronize with each
other and a synchronization unit (sold separately) is required. When cameras
capture frames in an unsynchronized fashion, measurements of the same marker
taken by different cameras are not simultaneous. This is, however, not a
significant drawback for most applications. If synchronization is required, there
are both HW and software solutions available. Please contact ClaroNav Inc.for
further information.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 35

Software
The following diagram shows the various software modules delivered as part of
the MicronTracker system and their dependency hierarchy. Modules below the
double line are supplied as executable libraries, and above it as example source
code.

PC, laptop or single-board

computer
Demo Applications

a Language wrappers (optional)

o
MTC library: C API

r
FlyCapture/Digiclops libraries

Camera firmware
Bumblebee driver
gle-board computer

w
re
6.4.1 Firmware
The camera firmware runs on a dedicated processor inside the camera and
communicates with the computer over the IEEE-1394 bus. The firmware controls
the sensing and communications hardware and is critical to accurate performance
of the product. Firmware version names contain 4 numbers
(major.minor.type.revision). The camera is calibrated for a specific firmware
version and updates to the firmware may require the camera to be recalibrated.

6.4.2 Bumblebee driver

The camera driver is provided by Point Grey Research. This driver depends on the
IEEE-1394 driver, which is part of the operating system installation.

6.4.3 FlyCapture/ Libraries

Higher level interfaces to the camera control functions, provided by Point Grey
Research.

6.4.4 MTC Library

MTC (MicronTracker’s C language interface) implements all the identification
and pinpointing algorithms and provides a complete object-oriented application
programming interface (API) for:

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 36

 Detecting and starting up the cameras connected to the computer, then loading
their calibration files;
 Loading Marker templates from files;
 Capturing and processing camera frames on demand to report on detected
markers and their poses;
 Creating or modifying marker templates;
 Customizing many aspects of the measurement operations to fit specific
application needs.
Applications written in any language may call MTC directly to take full advantage
of the MicronTracker’s capabilities. MTC completely encapsulates and abstracts
the lower-level camera control functions, making it unnecessary for the
application to make calls to any of the lower-level modules.
MTC version names contain 3 numbers: major.minor.revision. The major version
number, currently 2, remains fixed as long as backwards compatibility is
maintained, usually over a number of years. The minor version number
increments whenever new features are added, usually every few months. The
revision, or “patch”, number increments with versions containing new bug fixes,
but no API changes or feature additions.

6.4.5 Language wrappers

While C libraries can be interfaced directly from any language, programmers
developing applications in higher-level languages, such as C++, Visual Basic or
Python, find it easier to interface to the MicronTracker using a “native” API.
Since MTC is object-oriented, it is relatively easy and straightforward to write a
thin “wrapper” around the MTC API, making its objects appear as native objects
in the chosen language.
ClaroNav provides a number of example language wrappers, in source code
format, which can be used as are, or as examples to assist in developing a wrapper
for other languages of the developer’s preference. This code is neither
documented nor thoroughly tested. Furthermore, it is not guaranteed to provide a
full coverage of MTC functionality.

6.4.6 Demo Applications

A number of different demonstration applications are provided in source code
format to help people understand how to optimally use MicronTracker: They
include:
 MTDemoNet – an extensive demonstration of product capabilities, written in
Visual Studio .Net 2008. The demonstration project includes a C++ wrapper for
the MTC SDK, a VB.net demo and a C# demo.
 MTSimpleAppC – a simple, short, C console program that demonstrates the
basic MTC functionality.
 MTDemoVB6 – an extensive demonstration of product capabilities, written in
Visual Basic 6.
 MTDemoCPP – a multi platform limited demo that supports both Windows
and Linux environments and allows performing the basic MicronTracker tasks.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 37

6.5 General-Purpose digital Input/Output (GPIO)

The MicronTracker Sx60, Hx40, Hx60 and H3-60 cameras come with a 12-pin
GPIO connector on the back that allows the camera to interact with external
devices. It can be configured to trigger on an external electrical signal or produce
a similar signal that allows devices external to the camera to be synchronized to
the camera. To configure the GPIO pins, consult the camera’s “Frame
control/access” section of the MTC documentation. The following pictures show
the GPIO connector on the back of the Sx60 camera and its circular push-pull
coupling connector.

The GPIO pins are TTL 3.3V pins. Inputs can be configured to accept external
trigger signals. When configured as inputs, the pins are internally pulled high
using weak pull-up resistors to allow easy triggering of the camera by simply
shorting the pin to ground (GND). The inputs are protected from both over and
under voltage. It is recommended, however, that they only be connected to 5V or
3.3V digital logic signals. Outputs can be configured to send an output signal or
strobe pulse. When configured as outputs, each line can sink 10mA of current.

NOTE: To use the RS232 functionality, a level converter must be used to convert
the TTL digital logic levels to RS232 voltage levels. See B&B Electronics
(http://www.bb-elec.com/) P/N: 232LPTTL for an example.

NOTE: The GPIO connector shipped with the MicronTracker is only an example
of the type of connector required. To maintain EMC certification as a medical
device, any application that make use of the GPIO port should make sure that a
non conductive shell shall be present on the GPIO connector.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 38

7 Accuracy Performance
7.1 Reporting Terms
Accuracy, or trueness, is the closeness of measured values to their expected
quantity as defined by some standard. The difference between the true value and
the value reported by the instrument is called the measurement error. In
measuring a spatial position, the measurement error is a 3-vector (dx, dy, dz),
where each component is the difference between the measured position and the
true position projected onto the corresponding axis. The error value is often
reduced to a single number, which is the Euclidean length of the error vector, i.e.,
(dx2 +dy2+dz2).
Due to a variety of uncontrolled factors, repeated measurements of the same
quantity under different conditions generate slightly different values. It is,
therefore, common to take a statistical view of accuracy, describing it in terms of
error distribution around the true value.
A number of statistical values may be used to quantify error distribution in a
single number. The maximum error (or its close relative, the median error) is
easy to define, calculate and understand. Unfortunately, it is not a useful term in
characterizing measurement performance, since its value critically depends on the
number of samples taken and, due to random factors, may vary widely from one
experiment to another unless the number of samples is very large, sometimes
impractically so. The average error is also easy to define and calculate. It does
not, however, reflect well enough how “safe” it is to rely on the reported value. In
averaging, small errors are assigned the same importance as large errors, while
users often consider large errors more important than small ones. The average
error, therefore, tends to generate lower values than what users feel is the “true”
inaccuracy of the instrument.
Perhaps the most useful statistical error quantification is the % confidence
interval (CI), defined as the value below which a specified % of the samples fall.
For example, the value reported for the 95% confidence interval (95%CI) implies
that only 5% of the samples had higher error values. Unfortunately, CI values
have not been widely reported in the past, and there is no consensus regarding
what specific interval to use.
The most frequently used measure of spatial measurement accuracy is the Root
Mean Squared Error (RMSE), calculated by averaging the squares of the
individual error values, then taking the square root of the average. If all the errors
have the same magnitude, RMSE generates the same value as the average error.
When some of the errors are larger than others, however, they are given more
weight, resulting in the RMSE being larger than the average.
In the many experiments conducted to measure the accuracy of MicronTracker
products we found the ratios between RMSE and 95%CI to be very similar across
experiments:
95%CI  1.85 x RMSE

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 39

From publications, this ratio appears to be approximately valid in other optical

tracking products as well. Since 95%CI and RMSE are closely related and RMSE
has been widely used in the past, the accuracy characteristics of the
MicronTracker are specified using RMSE.
When the accurate value is not known, it is often useful to report on the
repeatability (or precision) of measurements in terms of their standard deviation
(STD). Standard deviation is identical to RMSE when the average is the true
value, i.e., it is calculated by taking the square root of the average of the squares
of deviations from the average value. When the average is off the true value by
distance d, the following relationship applies (assuming a normal error
distribution): RMSE = (d 2+STD2)

7.2 Application-independent accuracy reporting

The end-user of the application-specific system incorporating a MicronTracker
product is interested in the measurement accuracy of the overall system under
typical usage conditions. This accuracy will be affected by various factors, of
which many are either outside the scope of the measurements MicronTracker
provides, or would vary widely from one application to another. In particular,
system integrators may vary the design of markers and the tools/objects to which
they are attached.
Considering that
 Different marker designs would result in varying marker pose accuracy
characteristics; and
 Other vendors report on accuracy of single target position measurement;
ClaroNav Inc.has decided to characterize MicronTracker’s accuracy performance
by referring to the 3D position of a single target (Xpoint). While single target
positions are not typically used directly by applications, their measurement drives
the computation of marker poses and can, therefore, be used to model and predict
the marker pose measurement accuracy for a given marker design (see 10.5.2).

7.3 Factors affecting accuracy

The causes for measurement errors in optical pose sensors are numerous and their
effects are often complex and interdependent. It is useful to divide them into 3
categories: (1) calibration (2) drift and (3) jitter. Calibration errors are
introduced in the camera calibration process at the factory, and remain constant
for a long time unless the physical structure of the camera changes, e.g., due to a
physical shock. Drift errors are introduced by variations in the operating
environment, e.g., temperature, lighting, or marker position/orientation. Jitter is a
momentary deviation caused by random optical or electrical noise in the image
capture and analog-to-digital conversion circuitry.

7.4 Calibration
During calibration, a sequence of steps is executed to extract individual camera
characteristics, which are later used in converting the input pixel values into 3D
target positions. In the most critical calibration step coefficients are extracted for
the equations that convert pixel locations in the image into the corresponding

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 40

projection rays in camera coordinate space (see 2.4). To obtain those coefficients,
a 3D grid of targets at nominal spatial relationships to each other is presented to
the camera, and the projection equations’ coefficients are tuned to obtain the
closest match possible between the nominal spatial positions of the targets and the
equation representing the location of their projections in each image. Depending
on the model, 10,000-50,000 grid positions are used in the calibration of each
MicronTracker camera. Thus, many grid positions contribute to the calculation of
each projection coefficient. The effects of drift and jitter are reduced by
controlling the environment in which the calibration is made and by averaging the
results of repeated measurements of the same grid position.
Errors may be introduced during the calibration itself as a result of:
 Residual drift and jitter errors;
 Interpolation and grid registration errors;
 Deviations between nominal target grid positions and true positions.
Due to the large amount of redundancy in the process, the accumulated effect of
these errors expresses itself as a vector between each nominal grid position and its
position as calculated from the pairs of images acquired during the calibration.
The calibration accuracy is calculated from the magnitude of these error vectors.
The RMSE of the calibration accuracy has been adopted by suppliers of
competing pose tracking products and is, therefore, particularly useful in obtaining
a simple comparative accuracy yardstick.
A detailed record of the calibration process, including the calibration error
statistics and the range of coordinates over which it was measured, is stored in the
calibration file of each camera, and can be easily obtained through the API (see
8.3). ClaroNav Inc.guarantees that the calibration error of each camera shipped is
lower than specified.
Grid registration errors are minimal at the center of the volume over which the
calibration is performed, increasing towards its edges and beyond. To alert users
to the degraded accuracy, an “outside calibrated volume” hazard warning is
appended to the reported marker’s pose when the depth (z coordinate) exceeds the
farthest end of the FOM (1200/1150/2000/2400mm for H40,Hx40/S60,
Sx60/H60,Hx60/H3-60 models).
The specified maximum calibration error for MicronTracker is
0.20mm/0.25mm/0.35mm/0.20mm RMS for H40, Hx40, Sx60/H60, Hx60/H3-
60. The actual calibration error for a specific camera is recorded in the calibration
file header and can be viewed using the demonstration application (<Calibration
info> in the <Display> menu), or accessed programmatically through the MTC
API (Camera_CalibrationInfo or Camera_CalibrationFileInfo).

7.4.1 Calibration refinement with R-Fine

The MicronTracker camera is specifically designed to preserve the high accuracy
of its calibration performed at the factory. But, like any other optical pose tracker,
it can suffer from undesired hardware changes for a variety of reasons, such as
environmental stresses or rough handling. These modifications, however slight,
may in time render the camera’s factory calibration less accurate than originally
specified.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 41

To address the issue of a potential degradation of calibration accuracy, R-Fine

software application has been added to the MicronTracker kit. R-Fine provides a
fast and intuitive method to evaluate the camera’s calibration accuracy and the
possibility to refine the calibration, if required. Calibration refinement in the field
is a convenient alternative to shipping the camera back for full re-calibration. The
R-Fine procedure is recommended to be performed every few months or if the
camera has been subjected to physical stresses.
R-Fine software detects a specially designed
AccBar marker, which is supplied as part of
the MicronTracker kit. The software provides
visual guidance to place the AccBar at specific
positions and orientations in the camera’s
view. The goal is to collect a representative set
of AccBar samples, detected over the camera’s
FOM. Each sample contains an estimate of the
AccBar’s length, computed from its image
projections using the camera’s calibrated
projection rays.
The variability of the AccBar’s length (measured as its STD) over the camera’s
FOM is used to quantify a characteristic non-rigid distortion of the camera’s
calibrated space due to likely physical changes in the hardware. R-Fine uses the
collected samples to optimally adjust the camera’s calibration model in order to
lower the length variability and thus eliminate the distortion effect. Typical
variability results produced by an accurate calibration are considered to be within
0.20-0.30 mm range for Hx40, 0.25-0.35 mm range for Sx60 and 0.35-0.40 mm
range for Hx60 and 0.20-0.35 mm for H3-60.
In addition to the length variability, R-Fine reports a projection ray convergence
value before and after the refinement. Ray convergence describes the precision of
intersection of the camera’s projection rays, where they meet to form a 3D point
from its left and right image projections. Ray convergence is expressed as an RMS
shortest distance, or intersection gap, between the left and right rays
corresponding to a triangulated point. The RMS is computed over all pairs of
projection rays associated with the collected AccBar samples. The lower this
value, the closer the calibration resembles an ideal camera model, in which the
rays intersect at unique points in space.
NOTE: Because accuracy requirements are application-dependent,
MicronTracker users are encouraged to apply their own methods of evaluating
the camera’s accuracy in combination with the R-Fine protocol. Please contact us
if the camera’s tracking accuracy remains unsatisfactory after the refinement
procedure.

7.5 Drift Factors

Much effort in the development of the MicronTracker has been invested in
detecting, characterizing and mitigating drift errors. As a result, MicronTracker is
largely unaffected by external measurement conditions. In some cases, where the
drift is hard or impossible to eliminate, a decision was made to implement hazard
condition detection and associate a drift hazard warning with the affected

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 42

measurements. The following table summarizes how various potential drift factors
impact the performance of the MicronTracker:

Drift factor Impact

Rapid internal A thermal instability hazard is added to
temperature change in measurements until camera becomes thermally
camera stable (15 minutes for binocular models and 30
minutes for H3-60 model following power-up or 2
minutes following activation, whichever is later).
Ambient temperature Camera automatically corrects thermal drift within
operating temperature range (18-30°C or 66-86°F).
A hazard warning is issued when temperature falls
outside the range.
Atmospheric pressure Insignificant
Light intensity or Insignificant
direction
Light type/spectrum Insignificant when the light coolness property (see
next) is correctly set by the user. Incorrect setting
may affect the Z coordinate of measured positions
by up to 0.25% (although error rarely exceeds
0.1% in most common in-doors circumstances).
Xpoint rotation and tilt Insignificant*.
angle relative to camera
Image contrast around Insignificant.
Xpoint
Partial occlusion or Insignificant
staining of Xpoint
Uneven lighting on Small drift may result when part of an Xpoint is
Xpoint shadowed at certain orientations. This condition is
detected and a “Shadow over XP” hazard is
appended to the marker’s measurement.
Motion blur Insignificant.
Power supply voltage Insignificant.
variations
Camera mount and Insignificant
orientation vs. gravity
Camera exposure settings Insignificant
*) While Xpoint orientation by itself does not cause drift, a small sampling drift is
introduced when the Xlines are aligned with the image X and Y axes.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 43

7.5.1 Light Coolness

The spectral contents of the light (the mix of component
colors) reflected off Xpoints affects the rays’ refraction path
through the lenses and, therefore, the exact location where
each feature point appears on the sensor. The relevant
property is called, in the MicronTracker’s API, the “light
coolness”, representing the overall balance between the
intensities of the light rays along the continuum between the
blue (cool) and red (hot) ends of the visible spectrum. The
more the balance tips towards the blue end, the light is
cooler. Light coolness is closely related to the “light
temperature” scale often used in photography, with higher
temperatures representing a cooler color mix. The chart on
the left shows the positions of commonly found light
sources on the light coolness scale, some of them with their light temperature
rating (CFL stands for “compact fluorescent light”).
The MicronTracker software contains correction algorithms that automatically
compensate for changes in light coolness, eliminating the drift that would
otherwise result. The internal coolness scale has been calibrated such that 0.0
represents incandescent light and 1.0 represents Daylight CFL, and the allowed
value range is –0.5 to 1.5.
Setting the light coolness correctly can be
done either automatically or manually.
Both methods are demonstrated in the
VB6 demo provided as part of the SDK.
The automatic method uses a “color
vector” (show in transparent green in the
figure on right) containing patches of red
and blue at known positions between its base and head. When it appears in the
FOM, the vector is identified by the application, e.g., by being the long vector of a
marker with a known name, and submitted to MTC to be used for auto-
adjustment. For reliable results, it is very important that the marker surface would
be highly non-reflective (matte), since light reflections negatively affect the
coolness calibration’s accuracy. It is also important to use long-lasting, stable,
colors which will not fade (such pigment-based dyes).
Automatic light coolness adjustment can be either performed once at the
beginning of a measurement session by flashing the “CoolCard” provided as part
of the MicronTracker kit, or continuously, by leaving the CoolCard visible in the
FOM, or by embedding a coolness tuning “color vector” within markers in use for
other purposes, such a coordinates reference marker. MTC supports the concept of
“color profiles”, identifying a specific combination of color patch sequence, ink,
paper and printer. Currently, only a single profile, associated with the CoolCards
issued by Claron, is supported. The .net demos identify the long vector of the first
facet of any marker whose name starts with the letters “cool” as a color vector.
Manually setting the light coolness is straightforward, and is demonstrated in the
“Options” dialog of the demo applications (see Section 8.10).

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 44

7.6 Jitter
Measurement jitter is the direct result of image noise. Image noise, in turn, is
affected by the camera’s gain setting. MicronTracker’s automatic exposure
algorithms have been designed to optimize exposure settings for minimal
measurement jitter. Proprietary jitter-filtering algorithms further reduce
measurement jitter when markers are momentarily static. The jitter filtering has no
negative impact on measurement accuracy or response lag.
The following table shows the Hx40 jitter’s standard deviation under a selection
of typical measurement conditions. In general, jitter for static targets, such as for
reference markers, is very low under all conditions. Jitter magnitude for moving
targets is proportional to the root of the gain factor and the distance.

Condition Sx60, STD Hx40, STD

mm mm
Static target, z=750mm, 150 Lux, GainF = 2.9 <0.01 <0.02
Static target, z=750mm, 500 Lux, GainF = 1.0 <0.01 <0.02
Moving target, z=750mm, 150 Lux, GainF = 2.9 0.11 0.22
Moving target, z=750mm, 500 Lux, GainF = 1.0 0.07 0.14

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 45

8 The VB.Net Demo

Application
MicronTracker ships with a few example applications written in Visual Basic
.Net, Visual C#, CPP and VB6 in both executable and source formats. By default
the application is installed under Program Files\ClaroNav\MicronTracker. A
shortcut to this application is placed on the desktop and it is also available through
Start  Programs  ClaroNav MicronTracker. The applications demonstrate
usage of most of the MicronTracker API, and enable accomplishing many simple
measurement operations. A user can register markers by showing them to the
camera, record distance and angle measurements between markers as well as
accuracy statistics, register multiple cameras to create a combined FOM, take
frame snapshots, and produce an AVI time-lapse video of a measurement session.
A user may also modify various camera settings and immediately view their
effect.
Most of the demo applications include dynamically-loaded library (DLL)
providing a “wrapper” around the native C API of MicronTracker (MTC.dll). For
instance MTInterfacedotnet.dll is used both by VB.net and Visual C# demo
applications and the VB6 demo application includes an ActiveX class library
“wrapper” around the MTC.dll. On overview of the API is provided in section
10.9.
The VB.net and Visual C# applications are the most developed demo applications.
The source code is installed under Program
Files\ClaroNav\MicronTracker\MTDemosNet. In the rest of this chapter the main
features of the VB.net demo application are discussed.

Snapshot of the MicronTracker VB.net Demo application

NOTE: The VB6 Demo application code is used internally by ClaroNav during
development and quality control, which sometimes requires access to low-level or
advanced features not supported as part of the public API. The development-only
parts of the code are conditionally compiled, depending on the value of the “dev”

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 46

pre-processor variable (i.e., they are bracketed between “#if dev then” and “#end
if”).

8.1 Camera Selection

On startup, the demo application automatically attempts to activate all the

MicronTracker cameras connected to the computer. Once activated, tracking and
display commence immediately. The activated cameras are listed by the
MicronTracker model name, serial number and resolution on the right side of the
menu bar (red arrows in the above figure). If there is more than one camera,
selection of the “current” camera is available by clicking on the camera’s listing in
the menu bar. The current camera is indicated by an arrow on the left side of its
listing.
The current camera selection determines the images shown and the marker pose
measurements. Pose measurements made by other cameras registered to the
current one is shown as if made by the current camera, even if the current camera
does not see them.

8.2 The Activate Menu

The Activate menu allows starting/stopping the data capture or the frame
processing. The frame update rate and the data capture status are displayed at the
top right of the images. <Externally Trigger> and < Strobe Signal> are available
for Sx60, Hx40, Hx60 and H3-60 cameras. Enabling <Externally trigger> will
stop the normal capturing mode and will put the camera in different mode. In this
mode the camera, does capturing and frame processing right after receiving an
external signal trough GPIO connector. The <Strobe Signal> doesn’t interfere the
normal execution of the application and just transmits electrical signals when the
camera captures new images. Toggling on <Background Process> makes the

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 47

MicronTracker grab and process frames automatically in a background thread and

retrieves the most recently processed pose data. This allows faster processing and
it increases the performance of the capture process about 30%. Toggling off
<Auto Exposure> enables the user to manually control the camera’s shutter and
gain settings using controls which appear on the right of the top image. The <High
Dynamics Range Mode> toggle enables/disables HDR mode for the camera, a
feature that enables the camera to detect markers is a wider range of illumination
like under OR light. The <Cam-cam registration> toggle enables/disables the
automatic update of the camera-to-camera registration transform whenever the
same marker is measured by multiple cameras. The <Marker Templates…>
selection invokes a dialog to register new markers and manage the MicronTracker
markers template database (see 8.6). The <Pose Recorder…> entry activates a
dialog to record pose measurements for statistical analysis (see 8.7). The
<Snapshots…> activates a dialog to save and restore complete measurement
frames (see 8.8). The <3D Tracer…> dialog supports creating an augmented-
reality overlay trace (see 8.9). The <Options…> dialog allows modifying some
general tracker settings (see 8.10). For convenience, activating most menu options
is also possible through keyboard accelerator keys.
NOTE: The automatic exposure controller needs a constant stream of frames.
Enabling the <Externally trigger> mode, particularly with a significant time
delay between two frames, may cause the auto exposure to perform incorrectly. It
is advised to use manual exposure when <Externally trigger> mode is enabled.

8.3 The Display Menu

The Display menu allows selecting various options for image

and overlay display. Each option can be toggled on or off.
The <Hist. Equalize> option activates a special algorithm (a
variation of histogram-equalization) to maintain a visually
consistent image despite the constant manipulation of the
exposure settings by the MicronTracker. This is a particularly
useful option when image overlays are being used to mark
locations and objects of interest (“augmented reality”).

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 48

Enable <Markers Position at ref Space> to display the pose data at the
Reference space. Both marker and the tooltip pose data will be displayed at the
reference space. By disabling this option it will display the pose data at the
“Camera-Space”.
When <Magnified tooltip> is activated and a marker with a corresponding tooltip
is detected, the regions around the tooltip in both left and right images are
magnified 4X (from the original video image), contrast-enhanced, and displayed
as insets within the video images. A cross mark shows the assumed location of the
tooltip, as extrapolated from the marker location. This provides an excellent
indication of the exact location actually being reported as being at the tip of the
tool, and provides a simple visual inspection to detect problems with tooltip
registration (see 8.6.3), marker attachment or tooltip step bending.
When the <Vectors> option is on, the vectors of identified facets are shown in
cyan (the longer vector, defining the X axis) and magenta (the shorter one).
Vectors that were not matched to a known facet are shown in yellow.
When exposure parameters are displayed, the demo also shows an estimate of the
amount of illumination, in Lux, falling on the brightest marker tracked.
The demo includes support for identifying a variable number of vectors used for
marking a variable number of locations, and not associated with any specific
marker. These locations are called “fiducials” and are assumed to be marked by
vectors with length of 10mm or less (this parameter can be programmatically
modified by changing the value of the FiducialVectorMaxLength property of
the main form). When the <Fiducials> option is turned on, such vectors are
marked in the image by a green circle and the letter “f”.
The calibration record of your camera can be viewed by selecting <Calibration
info>.
<Save left image…> saves the left (top) picture, exactly as it appears on the
screen (not as captured), compressed using JPEG, in a file.

8.4 The Video Recorder Menu

The Video Recorder menu demonstrates the creation of a time lapse AVI file
documenting a procedure in video. When recording is activated, the user is
prompted to select the output AVI file path and the video compressor to use. The
time interval between recorded frames is determined by the setting of the
tmrAviSaver.Interval property of the main form (500 milliseconds by
default). To minimize interruptions to the pose measurement cycle on slower
computers, an uncompressed format is recommended. A blinking red dot appears
at the top right of the top image when recording is in progress.
To eliminate image brightness oscillations during the video recording, you should
ensure that the <Hist. Equalize> option in the Display menu is checked.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 49

8.5 Help
<MTC API Documentation> opens a web browser to show the home page of the
API documentation. The <About…> dialog box provides information on version
number of the executable files being used, and allows access to complete system
information.

8.6 The Marker Templates Dialog

NOTE: For optimal performance, new marker templates should be registered only
when no hazards are associated with the measurements.

8.6.1 Registering a new marker template (single facet)

To register a new marker template:
 Select <Templates…> from the Activate menu (or press Ctrl+T). The
Templates dialog is displayed. A list of the currently registered marker or cloud
templates is shown. The number of facets in each marker template appears in
parentheses next to its name.
 To register a marker select the Marker
Definition tab and type the new marker’s
name in the Name field. The name needs
to be a valid MS-Windows file name, and
different from any other registered
template’s name. Otherwise, the <Sample
facet> button will be disabled.
 Make sure that no other unregistered
markers are visible to the camera. Place
the new marker in front of the camera such
that it will be fully seen by both lenses and
click on <Sample Facet> to toggle
sampling on. When exactly two
unidentified vectors are seen by the
camera, the “Collected” counter should
increment with each frame. Otherwise, the
reason a sample cannot be collected
appears below the “Collected” line. If the
marker contains short vectors that should
be ignored (often the case when two XPs
share B/W regions), you can set a
minimum vector length to be recognized.
Once you have a substantial number of
samples (>30) you can stop the recording
by clicking again on the <Sample Facet>
button to toggle it off. If you are unsure of
the quality of the sampling, you can click
on <Reset> to discard the samples collected so far.
 Click on <Accept & Save>. The marker is now in the database. Its name
should appear next to the marker in the upper (left) image whenever the marker is
in the FOM.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 50

After saving the template, the database is automatically checked to see if the new
marker’s facet is too similar to other facets in the database, given the current
setting of the template matching tolerance (see 12.3). If a
similarity warning appears, you can choose to reduce the tolerance (Options
menu), delete the similar marker(s) from
the database, delete, redesign and re-
register the new marker, or ignore the
warning (taking the risk of possible
misidentification of markers).
Markers are stored in text files (.ini
format) in the “Markers” folder, which is
automatically created in the directory
where the application’s executable resides.
As necessary, they can be manually edited,
e.g., by using “Notepad” or “Wordpad”.
They may also be copied to other
computers and used with any
MicronTracker model.
The Templates form also enables to define
a reference marker. Select the Marker
properties Tab and from the known
Templates list select a marker and then
click on “Set this as the reference” button.
From this moment the selected marker will
be recognized as the reference marker for
all other known markers.
By clicking on each known marker from
the list, it will be displaying the name of
reference marker for that particular
marker. Note that if you select the
reference marker from the list it will show
“Camera-Space” as a reference for the reference marker.
NOTE: to remove incompatibility problems, the decimal dot in numbers stored in
marker files represented by a period, regardless of the OS regional number
format settings. Commas inserted manually would be interpreted as dots when the
marker files are loaded.

8.6.2 Registering a multi-facet marker template

When the marker contains more than one facet, start by registering the marker
with just a single facet as described in the previous section (8.6.1). If the multiple
facets face in the same direction, you may need to cover the centers of Xpoints of
all facets other than the one you are registering (e.g., by using the sticky part of a
post-it note). If the facets are facing in different directions, make sure only one of
them is facing the camera during the sampling step.
To add facets to the marker’s template:

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 51

 Make sure there is no thermal or other warning message displayed by the

application.
 In order to get the best registration accuracy, it is recommended to disable
Jitter filter, small marker detection and Template Based Warm-up
Correction from <Settings…> in the Activate menu. Also place the marker at
about 60cm far from the camera as close as to the middle of FOM.
 Make sure the template’s name appears in the “Name” box, and check
<Additional facet> on.
 Position the marker (or uncover hidden Xpoints) such that both at least one
registered facet and the facet you are adding are seen by the camera concurrently.
 Click the <Sample Facet> button to toggle sampling on.
 Collect samples of the additional facet together with the known facet (so that
the MicronTracker software will register their relative spatial positions). More
samples lead to better accuracy (it is recommended to collect ~400 samples while
marker is static).
 Once a sufficient number of samples has been collected, click the <Sample
Facet> button to toggle it off, then click <Accept and Save>. The facet count in
parenthesis next to the marker’s name in the list should increment by one.
Repeat the above steps for each additional facet.
In some cases the additional facet is placed such that it cannot be observed by the
camera concurrently with any previously recorded facet. For example, this is the
case when the marker has only two facets oriented back-to-back. Several options
are available for handling this case:
 Use an additional temporary facet during the template registration process.
That facet should be positioned such that it can be observed together with both the
registered and the unregistered facets. It is then registered relative to the first facet
and subsequently used to register the additional facet. After the registration has
been performed, this additional facet can be removed and ignored (you can
manually edit the template file to remove it, or make sure the same facet it not
registered with any other template).
 Use a multi-camera setting where the two facets can be simultaneously
observed by two or more mutually-registered cameras.
 Treat each facet as a separate marker, and make the application aware that
they refer to the same object, e.g. by assigning them marker names conforming to
some convention (eg. [objectname].L and [objectname].R for left and right facets
correspondingly).

8.6.2.1 Verifying accuracy of Multi-facet marker registration

Steps below are recommended to be taken in order to verify accuracy of multi-
facet marker registration.
 Place the marker at 60cm far from the camera
 Select <Pose Recorder> from the Activate menu and make sure the “marker
Origin position” is checked
 Collect 200 samples for each of the following scenarios:
a. While both facets are visible

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 52

b. without changing the experiment scenario cover a portion of facet #1

c. uncover the facet #1 and cover facet #2 and collect additional 200
samples
d. at the end uncover the both facets and collect another 200 samples.
 Save the recorded pose data and import them to the MS Excel to draw a
graph of Z pose data. If the registration was done correctly then there
shouldn’t be any jump in the graph when switching between different
combinations of facets. In fact the first and last 200 samples are taken under
exact same conditions and they should have exact same average Z value. If
the average is different then the experiment is not done correctly.
 Following diagram shows the results of a properly registered 2 facets marker.

8.6.3 Registering a tool tip

The demo provides a simple method for registering the position and orientation of
the tip of a tool, such as a pointer, simply by placing it at the origin of another
marker with a known name. An example of such marker, named “Ttblock” (for
Tool-Tip calibration Block) is provided as part of the MicronTracker SDK. Once
a tooltip is registered, its position is reported on the screen and can be recorded
using the Pose Recorder dialog. Furthermore, an arrow representing the predicted
location and orientation of the tip in camera space is overlaid on the video images,
allowing verification of the tip’s calibration procedure.
The picture below shows a suggested setup for performing the tooltip registration
using the example Ttblock marker shipped with your MicronTracker unit. A sheet
of paper covers the second marker on the plate, and a heavy notebook or book is
placed in front of the plate to prevent it from sliding when pressure is applied
against the small notch at the Ttblock’s origin. For maximum accuracy, the plate
should be placed at a distance of 40-50cm from the camera.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 53

NOTE: The mechanical structure of the Ttblock example does not provide highly
accurate results (typical errors can be up to 1mm in position and 20 degrees in
orientation). For more accurate results you can either order pre-calibrated tools
from Claron, or use a more robust and accurate tip calibration setup/procedure of
your own.
To perform the registration, ensure that only the tool’s marker and the “Ttblock”
marker are detected. Place the tip of the tool in the small notch on the board,
touching the origin point of Ttblock (the middle of the long vector), with the tip’s
stem at an elevation of 90 degrees directly above the orientation line pointing at
the marker origin. Click on the <sample tooltip> button to begin sampling. Once
a sufficient number of samples was collected (e.g., 50), click on the button again
to complete the sampling and save the result by clicking on <Accept and Save>.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 54

You should now observe the tooltip arrow overlaid on the video images. The tip
of the arrow points at the positive Z direction in the tooltip coordinate system, and
its side lobes lie in the tip’s XZ plane.

8.6.4 Registering a new marker template using UniTracker

To register a new marker template using UniTracker:
 Select <Marker Templates…> from the Activate menu (or press Ctrl+T).
The Marker Templates dialog is displayed. A list of the currently registered
marker templates is shown. The number of facets in each template appears in
parentheses next to its name.
 Type the new marker’s name in the Name field. The name needs to be a valid
MS-Windows file name, and different from any other registered template’s name.
Otherwise, the “Sample facet” button is disabled.
 Make sure that no other unregistered markers are visible to the camera. Mount
the new tool to the TTCalib marker rigidly and place them in front of the camera
such that it will be fully seen by the camera and click on <Sample Facet> to
toggle sampling on. Note that while collecting samples both the camera and the
new marker and TTCalib must be in static condition. When exactly two
unidentified vectors and the TTCalib marker are seen by the camera, the
“Collected” counter should increment with each frame. Otherwise, the reason a
sample cannot be collected appears below the “Collected” line. Once you have a
substantial number of samples (>50) you can stop the recording by clicking again
on the <Sample Facet> button to toggle it off. If you are unsure of the quality of
the sampling, you can click on <Reset> to discard the samples collected so far.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 55

 Without changing the relative position of the new tool with respect to the
camera, reposition the camera and collect another set of samples by clicking on
<Sample Facet> to toggle sampling on.

 Click on <Accept & Save>. The marker is now in the database. Its name
should appear next to the marker in the upper (left) image whenever the marker is
in the FOM.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 56

After saving the template, the database is automatically checked to see if the new
marker’s facet is too similar to other facets in the database, given the current
setting of the template matching tolerance (see 12.3). If a similarity warning
appears, you can choose to reduce the tolerance (Options menu), delete the similar
marker(s) from the database, delete, redesign and re-register the new marker, or
ignore the warning (taking the risk of possible misidentification of markers).
Markers are stored in text files (.ini format) in the “Markers” folder, which is
automatically created in the directory where the application’s executable resides.
As necessary, they can be manually edited, e.g., by using “Notepad” or
“Wordpad”. They may also be copied to other computers and used with any
MicronTracker model.

8.6.5 Registering a new Clould of Points

Cloud of points (COP) is a set of Xpoints with 3 or more Xpoints. Each individual
Xpoints in the COP has a unique index. Partial tracking of the COP is only
possible if at least 3 Xpoints are visible to the camera. To register a new Cloud of
points:
 Select <Templates…> from the Activate menu (or press Ctrl+T). The
Templates dialog is displayed. Select the “COP” tab. A list of the currently
registered Markers and COPs templates is shown.
 Type the new COP’s name in the Name field. The name needs to be a valid
MS-Windows file name, and different from any other registered template’s name.
 Place the Xpoints in the FOM, the camera and COPs must be static.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 57

Click on the “COP” button. Once enough samples are collected the Xpoints will
turn into green. Click again on the “COP” to toggle it off. Now the “Accept &
Save” button should be enabled. Click on it to save the template file.

You may need to close and run the application. From the activate menu select
“Clouds Processing” and from the Display menu select “Display Clouds”. The
registered cloud should be detected and the pose data will be displayed on the
screen.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 58

Once the COP gets detected, a COP object will be available with the following
information:
a. The name of the COP
b. Collection of Xpoints objects in the COP and their corresponding indexes and
3D positions in the Camera space. And information to determine whether
they have been visible to the camera or not.

8.6.6 Renaming and Deleting Markers

To delete or rename a marker, simply select the marker by clicking on its name in
the list, then click on the <Delete> or <Rename> button.

8.6.7 Filtering by known vector lengths in C# Demo

Depending on the design of a marker, there can sometimes be many extraneous
vectors identified by MicronTracker which make registering the marker difficult.
It is useful to be able to filter out any vectors that do not fall within a specific
tolerance of a marker design.
Currently only available in the C# demo, there is
an option to make this filtering of vectors very
straightforward. In the application folder a user
will see a text file called
KnownVectorTemplates.ini. In this file, the user
can specify the names of their markers and also
the lengths of all the vectors in this marker. The
default configuration shown below, shows how
one might define the TTBlock for this process:

[MarkerNames]
Count=1
Marker0=TTBlock

[TTBlock]
Tolerance=0.75
VectorCount=2
Vector0=90.0
Vector1=86.5

The file starts with the header “[MarkerNames]”.

Within this section, you can specify “Count”, the
number of markers you wish to define vector
templates for. Following count are the names of
all the markers you wish to define specified by
fields named “Marker#”.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 59

Each marker then has its own section, where fields “Tolerance”, “VectorCount”,
and fields for each vector length named “Vector#” are specified.
After loading the marker template dialog, an entry for each specified marker will
be listed in the drop down menu. By selecting the desired marker, the name will
automatically populate in the marker name text box. You can then proceed with
the normal registration process and the vectors lengths will be filtered
accordingly.

8.7 The Pose Recorder

The pose recorder’s purpose is to allow users to
conveniently export a large number of pose
measurements to a file for off-line analysis. The
format in which the measurements are exported
allows the file to be easily imported into
spreadsheet programs to perform various
calculations and statistical analyses on the
measurements. To open the file in MS-Excel, for
example, right-click on the file’s name, then
select Open With  Microsoft Office Excel (or
open the file in Excel).
The data at each sample is collected in an internal
record. Each data item in the record can
optionally appear in the formatted text by
toggling it on/off using the provided checkboxes.

NOTE: The state of the checkboxes has no effect

on what is being collected, only on what is being
written to the output file. After the data was
collected, you can save it multiple times to
different files with different options
enabled/disabled.

Other than the various automatic and manual

options controlling the data capture timing, it is
possible to use the <by file> toggle to enable
triggering record capture from another program
with access to the computer’s file system (i.e., a
local
program or on another computer in a local area network with proper access
permissions). When placed in the “by file” mode, the demo program checks for
the existence of the file “C:\Mtmeasure”. When it exists, a record is captured, the
file is erased to signal completion, and the program waits for it to appear again.
The frequency in which the file’s existence is checked is controlled by
tmrCheckFile.Interval property of frmPositionRecorder. You can also

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 60

easily change the name and/or location of the file by modifying the function
tmrCheckFile_Timer.
When <2 markers> statistics is chosen, the program shows distance and angle
statistics relating to records containing exactly two markers (only). Otherwise, it
shows statistics relating to the tooltip position of the first marker in each record
(which is identical to the marker’s origin position if a tooltip is not registered for
that marker).

8.8 The Snapshots Dialog

One or more snapshots of measurement frames (left and

right image uncompressed) can be recorded by clicking
the <Snap frame> button. Each snapshot is appended to
the end of the snapshots list. Using the <Save…>
button, the list may be saved in a special “stereo
snapshots” files (.ssnap extension), from which it can
retrieved later by using <Load new…> or <Load
append…>. Scrolling through the list, using the slider
or the frame number box above it, feeds the snapshots to
the MicronTracker’s interface software as if they were
newly captured frames from the camera. Regular data
capture is disabled, and can be enabled again by clicking
the <Data Capture> option in the Activate menu (or by
pressing Ctrl+D).
NOTE: correctly performing measurements from
snapshots requires that the calibration file loaded will
be the same one used when the snapshots were taken.
Calibration files are unique to each individual camera. A warning is, therefore,
issued when a snapshots file is loaded while the currently connected camera is
different than the one used when the snapshots were taken.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 61

8.9 The 3D Tracer Dialog

The 3D Tracer feature demonstrates
how to use the MicronTracker’s API
to generate and display 3D geometry
projected and overlaid on the image
while “anchored” to a (reference)
marker. The demonstration uses
manual tracing to generate the
geometry. The manual tracing may
be performed, for example, to mark
the location of visible boundary or
feature points of an object assumed
to be fixed in space relative to the
reference marker during
measurements. If the object
accidentally shifts, a mismatch
between the trace and the locations it
marks would appear.
To trace, you need to enter the name of the reference marker, the name of the
tracer marker and click on the <Trace> button. When both markers are detected,
line segments are added to the trace. The trace, if visible, is shown overlaid over
both the left and the right pictures.
If no reference marker is entered, the segments are recorded in the camera space,
and will thus be fixed to the camera. Otherwise, the line is anchored to the
reference marker and would move with it when the marker or the camera is
moved.
When tracing is activated, it is often convenient to start/stop the tracing by
hiding/exposing the tracer marker.
To better examine of the match between underlying feature locations in the image
and the trace, the region around the center of the trace may be shown magnified
by checking the <Show magnified> box.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 62

8.10 The Options Dialog

This form allows modifying various
attributes that control the performance of the
MicronTracker. Entering a value for an
option other than its currently set one causes
it to be highlighted in red. To actually set the
MicronTracker’s value, click the <Apply>
button.
The meanings of most of the attributes
controlled by this dialog are described in
detail in sections 12 and 7.5.1 (light
coolness).
The Properties tab provides options to
control Marker and Cloud detection
behavior. For instance the jitter filter
provides measurement noise reduction for
static markers by averaging multiple
measurements taken at the same location. It
automatically shuts off at the slightest
movement, and does not, therefore, have an
adverse effect on dynamic measurements. It
should, therefore, remain enabled during
normal operation. You may want to disable
it temporarily, however, in order to quantify
the jitter level during dynamic measurements
without actually having to move the marker
being measured.
Selecting Detect small markers option will
change the minimum Xpoint image
projection, required for detection, from 11 to
9 pixels. This option can be used to extend
the visibility of markers, enabling detection
of smaller Xpoint targets, or targets farther
away from the camera.
The Template Based Warm-up Correction
compensates measured pose data using
geometrical information stored in the
marker’s template file. If the marker is
damaged and the template file does not
correlate with the measured data,
MicronTracker will ignore correction and
will issue a warning message.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 63

Setting the number of maximum extrapolated

frames to a non-zero value can be used to
reduce flickering of tracked targets due to
momentary detection failure. If a marker is
no longer detected, its predicted location can
be used instead, extrapolated from its last
successfully detected position. The number
of extrapolated frames (ranging from 0 to 10)
is the number of consecutive frames in which
the extrapolated location of the undetected
marker will be reported. This option is most
effective for static or nearly static markers, as
rapidly moving targets may result in their
predicted locations lagging behind their
actual ones, if the number of extrapolated
frames is set too high.

The Cameras tab provides options to control

camera behavior like Shutter and Gain
limiter and strobe signal properties.
The Frame Rate attribute enables to grab
images with any desired frame rates. Its
default value is the camera’s maximum
supported frame rate. You may change the
frame rate either by moving the horizontal
scroll bar or by typing numerical values in
the corresponding text box.

The Coolness and HDR Coolness tabs

provide options to control coolness
parameter. Section 6.5.1 has more details
about coolness.
Enabling Auto Adjust HDR Min Lux allows
MicronTracker to automatically adjust the
Min value for the High Dynamic Range
mode. MicronTracker uses HDR Min
Detection value only if Auto Adjust HDR
Min Lux is disabled.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 64

9 The R-Fine Application

9.1 R-Fine Overview
The R-Fine application comes as part of the MicronTracker kit and can be used
for both verifying and restoring the accuracy of the camera’s calibration (see
7.4.1). The executable is installed by default under Program
Files\ClaroNav\MicronTracker. Its shortcut is placed on the desktop and is also
available through the Start menu.
R-Fine operates by detecting a special two-faceted AccBar marker, mounted on a
metal S-shaped tool. The tool must be placed by the user in a number of different
regions and orientations, as guided by the software using semi-transparent image
overlays. Once correctly positioned, the AccBar’s computed 3D coordinates are
sampled to be used for calibration accuracy analysis.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 65

9.2 Main screen

To correctly position the AccBar tool in the camera’s field of measurement, R-
Fine uses the following visual guides:

1. 3D arrow (at the center of the screen), pointing in the direction the
AccBar must be moved to reach its required location.
2. (Optional) outline of the FOM region inside which the AccBar must be
placed.
3. Virtual AccBar icon, showing how the tool must be oriented in the
camera’s view (aligned with either X, Y or Z axis of the camera).
Aligning the actual tool with its virtual image will correctly place the
AccBar inside the current sampling region.
4. A status bar in the bottom right corner, displaying the percentage of
AccBar samples collected so far.
5. A status message (bottom left corner) reporting the progress of sample
collection. The message can be one of the following:
o AccBar not found (the marker is not detected by MTC);
o Sampling disabled: reduced accuracy (a hazard warning is
issued);
o Low light (the operating environment is not sufficiently
illuminated);

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 66

o Move AccBar into region (the marker is outside of the sampling

region);
o Orient AccBar as shown (the marker is not oriented properly);
o Collecting samples: N (N is the number of samples collected in
the current region for the current orientation).
The remaining components of the R-Fine’s main screen, such as camera’s
information or hazard warnings, are identical to those of the VB6 Demo
application.
NOTE: R-Fine procedure should be performed regularly every 3 months in order
to maintain acceptable accuracy for camera’s calibration file.
NOTE: R-Fine does not collect samples while a hazard warning is issued by the
MTC, to avoid possible drift errors. This implies that the R-Fine procedure cannot
be performed while the camera is still warming up after being powered
NOTE: R-Fine has a built-in check to discard duplicate samples, i.e. a minimum
displacement between two consecutive samples is required. If the sampling has
slowed down or paused, move the bar slightly while maintaining its orientation
within the region.

9.3 The Application menu

The Application menu provides the option of restarting the R-Fine’s sample
collection procedure, or exiting the application.

9.4 The Display menu

Much of the selections in the Display menu are the same as those for the VB6
Demo (see 8.3). In addition, <Region Borders> option is provided to toggle
on or off the display of the 3D outline, highlighting the location in the FOM
where the AccBar must be placed.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 67

9.5 The Help menu

The <Documentation> opens a default web browser to the home page of the
R-Fine documentation. The <About…> dialog box shows version numbers of
the executable and DLL files, and allows access to complete system
information.

9.6 The Camera

Similarly to the VB6 Demo, on startup R-Fine automatically detects any

MicronTracker cameras connected to the computer via FireWire. However, R-
Fine allows only a single camera to be activated at a time of its execution
(otherwise an error message is issued). The serial number and resolution of the
connected camera are displayed on the right side of the menu bar.

9.7 The Results window

Once all of the required samples have been collected, R-Fine performs their
statistical analysis and automatically executes a refinement algorithm, which
adjusts the calibration to best fit the data. The improvement in accuracy resulting
from the refinement is measured as a percentage of a combined decrease in the
AccBar length variability and projection ray convergence values (see 7.4.1).
Upon successful refinement (resulting in a non-zero improvement), the application
provides a choice of permanently modifying the calibration by updating the
camera’s calibration file. If automatic saving is enabled, R-Fine will instead save
the improvement result, and show the Results dialog for information.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 68

The following steps are recommended depending on the improvement reported by

R-Fine:
 If improvement is above 10%, it is recommended to save the refinement
parameters to the calibration file and run R-Fine again. Continue running
R-Fine until an improvement less than 10% is achieved. If an
improvement greater than 10% occurs 3 times in a row, please contact
support.
 If improvement is below 10%, the refinements should not be saved. In
this case either there has not been a significant degradation in accuracy, or
a previous R-Fine refinement has sufficiently improved accuracy.
 In automatic mode, any improvement will be saved. The calibration file
will indicate whether the refinement correction applied is Verified, which
indicates that a result below 10% improvement has been saved.
Once saved, the modifications to the camera’s calibration will be visible to the
MTC library. A brief record of calibration refinement can be viewed by selecting
<Display/Calibration info> in the VB6 Demo:

If an attempt to refine the calibration failed, the user is provided with an option to
retry the procedure by collecting a new set of AccBar samples (this will restart R-
Fine), or exit the application. Refinement failure is usually an indication that the
camera’s calibration hasn’t yet undergone a significant degradation in accuracy.
This outcome should also be supported by a low measure of length variability
before and after the refinement (typical variability results for a recently calibrated

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 69

camera are in 0.25-0.35 mm range for Sx60, 0.20-0.30 mm range for H40 and
Hx40, 0.30-0.40 mm range for H60 and Hx60, 0.20-0.35 mm for H3-60).

9.8 The log file

During its execution, R-Fine generates a log file (serial#_mm_dd_yy.log), written
to the same directory as the executable. The log records some additional
information that can be used for more detailed calibration analysis as well as
troubleshooting. The contents of the log file include:

 date of the R-Fine procedure,

 camera’s calibration information, including its serial number and model,
 coordinates of FOM regions,
 number of collected samples, per region and per orientation,
 average AccBar length, computed over the sample set, before and after the
refinement,
 standard deviation of AccBar length, before and after the refinement,
 projection ray convergence (intersection “gap”), before and after the
refinement,
 length average and standard deviation values per orientation, before and
after the refinement,
 internal parameter values used by the refinement algorithm.

9.9 R-Fine Quick Guide

This section provides a quick reference for basic details on how to use R-Fine to
re-calibrate a camera. For a more detailed description of the application, refer to
the previous sections.
Upon opening the R-Fine application, the user is presented with a semi-
transparent image of the AccBar tool. The user must position and orient the tool as
shown on the screen.
 A Yellow guide is shown when the camera cannot detect the bar.
 A Red guide is shown when the bar is detected but not in the correct position.
 A Green guide is shown when the bar is positioned and oriented correctly.
Try to overlap AccBar with the guide

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 70

Sometimes it is required to move the bar towards or away from the camera

The bar must be oriented as depicted

In this orientation the bar is held pointing towards the camera

After all samples have been collected, the R-Fine application will determine
whether the camera is out of alignment or not. The application will automatically
update the calibration when there is improvement. If more than 10% improvement
is observed, we ask that you repeat the process until you achieve less than or equal

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 71

to 10% improvement. If the improvement is greater than 10% after 3 consecutive

attempts, please contact support. This is an indication that the camera is out of
alignment further than R-Fine can correct for.
Improvement is greater than 10%. Repeat R-Fine procedure.

Improvement is less than or equal to 10%. Refinement is successful.

Improvement is greater than 10% for 3 consecutive tries. Contact support.

© Copyright ClaroNav, Inc., 2003-2016

10 Marker Design
10.1 Introduction
A marker is made of one or more patterned surfaces rigidly attached to, or printed
directly on, an object, usually either a reference body or a tool of some sort.
Marker design is an art. While there are rules that limit, to some degree, what
designs will be identified and measured by the MicronTracker, there is much
room left for creativity in finding the best design for a particular object used in a
particular application setting. The purpose of this section is to guide the marker
design process. Terms used in this section are defined in section 2.5 above.

10.2 XP contrast
For best visibility and accuracy under typical lighting conditions, the pattern
making up each XP should be printed using the highest possible contrast between
the dark and white regions, at the highest possible resolution. The surface should
be as non-reflecting (matte) as possible. The brightness of the surface should not
be easily affected by contact with human hands. If it would be used for along
period, it needs to be easy to clean and, as necessary, sterilize.
NOTE: Non-reflective XPs with deep black and bright white regions not only
provide the best identification reliability, but also improve accuracy by ensuring
each region appears highly uniform in the video images.
Under intense or mixed lighting conditions, Xpoint contrast may be reduced to
allow its detection in a situation in which it may otherwise be over-exposed
(“washed-out”). For example, XPs printed in black on a 33% gray background
(1/3 the brightness of white) would still be detected concurrently with a white-
background marker even when it is illuminated by a light that is 15 times more
intense (see pictures below).

Uniform lighting

spotlight
 MicronTracker Developer’s Manual, MTC3.8 73

NOTE: Slightly larger black/gray Xpoint regions are required to obtain the same
maximum detection range as black/white Xpoint regions.

10.3 XP flatness
The surface on which XPs are printed should be flat. A slightly curved Xpoint
region would still be detected, but may produce position reports that will vary
somewhat depending on the orientation of the Xnormal relative to the camera.
This small drift may still be acceptable for some applications.

10.4 Xlines
The two Xlines do not need to be perfectly perpendicular to one another, but this
is desirable for maximum angular detection range.
To eliminate measurement errors resulting from incorrect determination of the
extent of the Xpoint region, Xlines should preferably terminate at a sharp angle.
This is especially important when two Xpoint regions are placed next to each
other at an angle.

Good Bad

! ! ?

If the neighboring Xpoints are a mirror image of each other, it is often useful to
have them share an Xline as shown below.

XLine sharing

10.4.1 Size and shape

Maximum Xpoint detection distance is linearly proportional to the radius of the
Xpoint’s region. This radius is defined as the shortest distance between the Xpoint
itself and the nearest point on the external edge of the uniformly white or black
regions creating it. Increasing Xpoint region size therefore increases the field of
measurement (FOM). While the shape with minimal area for a given radius is a
circle, it is perfectly acceptable for the Xpoint region to be square or any other
shape.
As a rule of thumb, the maximum detection distance of an Xpoint is roughly 100
times its radius. For example, a 5mm radius would allow detection up to 50cm,
and a 12mm radius would provide detection up to the far limit of the FOM.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 74

Xpoint position measurement errors are also

linearly proportional to the distance from the XPoint regions’ radii
camera. Beyond a certain distance, those errors
become larger than the marker template matching
tolerance (controlled by the property
TemplateMatchToleranceMM of the Markers
object, see 12.3) making marker identification
unreliable.
NOTE: While it may be tempting in some
situations to increase the FOM depth by using
larger XPs, keep in mind that at depths larger
than the specified FOM, measurement errors
grow rapidly, roughly proportionately to the Max detection distance 
square of the distance. 100 x radius

10.5 Facet Design

10.5.1 Design rules
To be recognized as a unique facet, the facet pattern design must comply with 4
simple rules:
 A facet contains exactly two vectors.
 The angle between the vectors is between 8 and 172 degrees.
 One of the facet’s vectors is longer than the other by more than twice the
value of Markers_TemplateMatchToleranceMM (see 12.3). The default
factory value of this property is 1mm, which implies a difference of more than
2mm is required).
 The position of at least one end of its two vectors (in the facet’s own
coordinates) needs to be different than the corresponding end position in all other
facets registered in the marker templates database by at least twice the value of
Markers_TemplateMatchToleranceMM.
NOTE: By definition, to form a vector, one of the Xlines of one Xpoint must be
aligned with one of the Xlines of the other Xpoint, with black facing black and
white facing white. The Xpoint regions at both ends should lie on the same plane.
The MicronTracker software uses a tolerance of 5 degrees between the Xline and
the line connecting the two Xpoints in recognizing Xline alignment in the Xpoints’
projection in an image.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 75

Vectors Not Vectors

NOTE: Two XPs sharing an Xline automatically form a vector. This vector,
however, does not necessarily have to participate in the facet to which the XPs
belong. The marker templates dialog of the VB demo provides a simple way to
ignore such vectors when registering a template by setting a minimal vector
length threshold (see 8.6.1).
The facet pose (coordinate transform) reported by the MicronTracker software is
the one that minimizes RMS error between the facet template positioned at the
reported pose and the actual measured positions of each Xpoint. Therefore, all else
being equal, the more Xpoints in a facet, the lower the overall pose measurement
error. Since two vectors can either share or not share a single Xpoint, the number
of Xpoints in a single facet is either 3 or 4. It is possible, however, to construct a
flat marker with any number of Xpoints higher than 3 (except 5) by combining
multiple facets into a multi-facet marker (see 10.6.2 below).

10.5.2 Tooltip accuracy

Due to the lever effect, the farther apart the Xpoints at the ends of a vector, the
lower the error in measuring the vector’s orientation in space, and, therefore, the
lower the error in extrapolating object positions of interest, such as a tool’s tip.
The drawing below demonstrates this effect by showing a tool tip position error
range for two different placements of Xpoints. To minimize extrapolation error,
one end of the longer vector should be placed as close to the tooltip as possible,
and the other as further away from it as possible.
Assuming the tool tip is approximately aligned with the long vector of the marker,
and the vector’s length is l (refer to the diagram below), and given a position RMS
error at each marker Xpoint e(f) and a distance d between f and the tooltip point tt,
the tooltip RMS error e(tt) is approximately:
e(tt)  e(f) + 1.5e(f)d/l

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 76

If d is small relative to l, the tip error is dominated by the Xpoint error (roughly,
calibration error + jitter, typically in the range 0.2mm-0.4mm RMS). If d is large
relative to l, the error is dominated by the ratio d/l. As a rule of thumb, to ensure
that RMS error at the tip is sub-millimeter, keep ld.

range of tool tip

position
extrapolated
from
measurements
Xpoint with position
measurement error margin
f tt

l d

c
a
m
10.5.3 Facet design alternatives e
r
10.5.3.1 For tooltip sensing a
When the marker’s purpose is to provide the tooltip sensing, and the orientation is
unimportant or secondary, two simple designs, shown below, should be
considered.

Narrow “L” design

tip

Narrow “X” design

tip

The L design provides the most efficient use of marker area, and is thus preferable
when minimizing marker size is paramount. Accuracy degrades quite rapidly

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 77

away from the X axis direction (the blue arrow), and it is thus very important to
place the marker such that the X axis will pass through, or very near, the tip.
The narrow X design requires a larger area at the front, but reduces tip position
RMS error, by 10-20% when the tip is placed along the X axis, and by up to 50%
when the tip is off-axis. It also improves the roll angle sensing accuracy. Note that
the angle between the two vectors needs to exceed 8 degrees.
10.5.3.2 For general pose sensing
When angular measurement accuracy is equally important in all directions, the
following two simple designs should be considered.

“L” design
“X” design

10.6 Multi-facet Marker

A marker may contain an arbitrary number of facets. Unlike vectors within a
single facet, different facets cannot share Xpoints (and, therefore, vectors).
There may be two good reasons for designing a marker with more than one facet:
 Increasing the detection angle range by placing facets in different orientations.
 Increasing pose measurement accuracy by increasing the number of Xpoints
contributing to the pose calculation.
NOTE: While Xpoints may be shared by vectors of the same facet, they cannot be
shared by vectors of different facets. Each facet requires its own set of Xpoints.

10.6.1 Increased detection angle

When the Xnormal is tilted away from aiming at the lens, the Xpoint region’s
projection on the corresponding sensor shrinks by the cosine of the tilt angle. As
the Xpoint region’s projected image size falls below the minimum needed for
detection (11 pixel units diameter), MicronTracker cannot continue tracking that
Xpoint. Once this happens to one of the facet’s Xpoints in one of the two sensors,
tracking of that facet it lost, limiting the angular detection range of that facet.
Adding more facets to the marker, each facing in a different direction, increases
the angular detection range to the combined angular ranges of all its facets. For
example, placing mirror images of a facet on the right and left sides of a tool
allows it to be used with the camera either on its left or on its right. Adding

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 78

another facet on top allows detection from above as well, providing the full
angular detection range needed in practice. The picture below shows two views of
an example of such an arrangement. The three facets contain 3 Xpoints each.

10.6.2 Increasing pose measurement accuracy

Due to the error distribution algorithm the MicronTracker uses in matching
measured XP positions to their corresponding marker template positions,
increasing the number of XPs tracked in a marker increases the marker’s pose
measurement accuracy. A single facet can contain a maximum of 4 XPs. By
placing multiple facets on the same surface it is possible to create surfaces with an
arbitrary number of XPs (except 5). This is especially useful when creating a
reference marker, which defines the coordinate system for other markers nearby.
Error distribution is approximately normal in shape. In theory, RMS error
magnitude should thus decrease proportionally to the square root of the number of
XPs, e.g., a position measurement derived from 9 XPs should be roughly twice as
accurate as one calculated from 3 XPs. This calculation, however, implicitly
assumes that all XPs have a similar influence on the position’s calculation. In
practice, some XPs have a much greater influence than others, and, therefore,
increasing the number of XPs may not necessarily have the expected effect. For
example, when extrapolating to a position away from the marker, e.g., a tool tip
position, XPs located farther away from the position, or at the ends of a vector
perpendicular to the extrapolation direction, have a smaller impact on that
measurement.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 79

10.7 Reference marker design

A reference marker is used to establish a coordinate system anchored to an object
being measured, so that tooltip measurements would be made relative to that
object, rather than relative to the camera. This eliminates the potentially large
measurement errors caused by either object or camera movements during a
measurement session. For example, attaching a reference marker to the head
during a sinus surgery procedure allows the surgeon to conveniently move the
head during the procedure without requiring a re-registration. Furthermore, if line-
of-sight interferences are detected during the procedure, the surgeon is free to
reposition the camera, again without the need for a re-registration.
When considering how to design a reference marker, it is useful to think of the
region of interest (ROI) for measurements (e.g., the sinuses in the abovementioned
example) as the “tooltip” and of the region where the reference would be attached
as the tool “handle”. The guidelines and formulas for tool marker design would
then similarly apply. For example the formula for computing accuracy at the
tooltip (see 10.5.2) could be used to compute accuracy at the region of interest.
Some general guidelines for obtaining maximum accuracy in a reference marker
design:
 Interpolation is much more accurate than extrapolation, so placing the XPs of
the reference marker around the ROI provides the highest possible accuracy.
 If placing XPs around the ROI is not possible:
o Try to place as many XPs as possible as close as possible to the ROI.
o At least two XPs should be placed as far away as possible from the
ROI and from each other to maximize angular accuracy.

10.8 Design examples

Below are shown 4 different patterns for a 2-part marker designed to be mounted
along the handle of a tool and provide maximum accuracy at the tip of the tool,
which is extended along the narrow dimension of the marker (see arrow). A and B
use two facets (8 and 6 XPs), while C and D use only a single facet (4 and 3 XPs).
The two vectors of each facet are marked with similar-colored arrow overlays.
Compared to design D, the smallest and simplest, RMS error at the tip of the other
designs was measured to be about 55%, 65% and 75% for A, B and C
correspondingly.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 80

Next are two more examples of multi-facet markers to be used to provide a

reference coordinate system for other markers. The top one is a circular marker
containing 12 Xpoints organized into three facets. The vector directions are
distributed such that accuracy would be nearly equal in all directions. When
visually registering this marker (see 8.6), initially partially cover the XPs of all the
facets except one, and then add the other two facets, one at a time, by
incrementally removing the covers. Set a vector length threshold to hide the short
vectors along the shared Xlines.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 81

The design below is intended for a marker to be placed to one side of the region in
which measurements are performed, and is, therefore, designed to provide
increased angular accuracy in the direction of that region.

4-facet
mark

designed
or impr
v
d

ni-di
ectional accuracy.

2-facets narrow marker

10.9 Materials
Paper printed using an inkjet or laser printer and glued to a flat surface is perfectly
suitable for prototyping and experimentation. For parts used in a clinical setting, it
is desirable to use a highly durable material, one that can be easily cleaned and
can withstand many cycles of steam sterilization. Such a material is, for example,
“0.008” hardcoated fine textured weatherable Polyester” available from Cellotape-
Landmark Label.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 82

Circular or over XP regions made of such material can be obtained from ClaroNav
under the trade name Duramarks. They can be fixated to a marker carrying surface
in an accurately repeatable manner by creating 1mm deep “pockets” in that
surface. Larger marker stickers can be custom printed on order by professional
label makers, such as Cellotape-Landmark.

10.10 To learn more

A good discussion of the math behind marker design can be found in the paper
“Designing Optically Tracked Instruments for Image-Guided Surgery”, by JB
West and CR Maurer, IEEE Transactions on Medical Imaging vol. 23 No. 5, May
2004

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 83

11 Programming Interface
11.1 Programming Model

The figure above shows the object classes used in controlling the sensor and
obtaining pose measurements. One object of each of three classes, Cameras,
Markers and Xpoints, is automatically created when the interface library starts
running. The application refers to those instances directly by their class name
without the use of an instance handle.
The Cameras object represents the array of MicronTracker cameras performing
pose measurements. Calling Cameras_AttachAvailableCameras establishes
the communication link with all the cameras physically connected to the computer
and loads the proper calibration file for each of the cameras. Handles to individual
camera objects representing attached cameras may be obtained using
Cameras_ItemGet. Each camera can then be individually accessed and
controlled. However, the most common operations relating to cameras can be
called collectively through the Cameras interface. Upon terminating the use of
cameras, Cameras_Detach must be called to signal the driver to disassociate
itself from the application process.
The Markers object represents all the markers being potentially tracked by all
cameras, i.e., markers with a registered template. Each marker is represented by a
separate Marker object, which, in turn, maintains a collection of Facet objects, one
per facet, in its template. Each Facet object maintains two Vector objects. A

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 84

Marker object also optionally maintains a Tooltip2MarkerXf Xform3D object

describing the pose of a tool-tip in the marker coordinates.
The Xpoints object represents an array of all the Xpoints being detected by all
cameras. Each Xpoint is represented by a separate Xpoint object and can be
individually accessed and provides 3D spatial position and all 2D information of
detected Xpoint in the left and right images.
The Collection class represents a variable-length 1-based array of numbers,
usually object handles. It was designed to be roughly compatible with the ActiveX
class of the same name. The Persistence class represents a collection of variables
(name-value pairs) stored in a file. The file format is with the used for .INI files in
MS-Windows, and uses plain text, which can be easily manually edited. The
StopWatch class provides a convenient and accurate way of measuring elapsed
time.
Marker templates are stored in Persistence files, typically one per marker. The
Markers object creates and maintains an array of marker objects, each
representing a template. Templates may be added individually to the collection by
the client, or a single method, Markers_LoadTemplates, may be used to load
all templates in a given directory.

There are two different measurement cycles, Markers Processing and Xpoints
processing:
1. Markers Processing: A measurement cycle for Markers Processing includes
the following steps:
 Calling Cameras_GrabFrame to transfer the latest frame to the MTC’s
buffers;
 Calling Markers_ProcessFrame to extract measurements from the
frame buffers;
 Calling Markers_IdentifiedMarkersGet to obtain a collection of
handles to the identified markers;
 For each identified marker in the collection, calling
Marker_Marker2CameraXfGet to set the attributes of an application-
owned Xform3D object to the pose of the marker. Optionally, the tooltip
pose, obtained via Marker_Tooltip2MarkerXfGet, may be
concatenated to the marker’s pose to obtain the pose of the tooltip in the
camera’s coordinates.
 Accessing the Xform3D attributes to perform application-specific
operations.
NOTE: To speed up frame grabbing and markers processing, MicronTracker can
also do all the computation in the background when enabled with
Markers_BackGroundProcessSet().
This will make MicronTracker grab and process frames automatically in a
background thread and return to the user the most recently processed pose data
when requested.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 85

2. Xpoints Processing: A measurement cycle for Xpoints Processing includes the

following steps:
 Calling Cameras_GrabFrame to transfer the latest frame to the MTC’s
buffers;
 Calling Xpoints_ProcessFrame to detect each individual Xpoints from
the frame buffers;
 Calling Xpoints_DetectedXPointsGet to obtain a collection of
handles to the detected Xpoints;
 For each detected Xpoints in the collection, calling
Xpoint_2DpositionGet and Xpoint_3DpositionGet provide 3D
and 2D position data of the Xpoint.
 Vectors which were recognized in the frame(s) but not identified as being
part of a marker (i.e., did not match any template), can also be obtained by
calling Markers_UnidentifiedVectorsGet.
NOTE: It is critical that Cameras_Detach is called before the application
program terminates. Otherwise, the camera driver may keep the application
process running for a few more minutes before releasing it, preventing access to
the camera by any other process. In an interactive development environment, such
as VB, terminating the application program without calling Cameras_Detach
would frequently crash the development environment process.

11.2 Detailed C API Description

MTC supports over 200 function calls, enabling the application to tightly control
and modify many aspects of MicronTracker’s hardware and software operations.
Those functions are documented in a set of hyperlinked web pages that can be
accessed from Windows task bar (Start  Programs  ClaroNav
MicronTracker SDK documentation) or from the VB6Demo menu (Help  MTC
API documentation). The “home” page is in {$MTHome}\doc\html\index.html,
where {$MTHome} stands for the path where the MicronTracker software was
installed (by default, C:\Program Files\ClaroNav\MicronTracker).

11.3 Source Code Examples

Following installation, the selected installation folder should contain a number of
demonstration applications, each in its own sub-folder:
 MT_SimpleDemoC: A barebones example that shows a basic of operations
for obtaining measurements from a single or a multi- camera set.
 MTDemoCPP: A multi-platform demo for both Win32, Linux and Mac OSX.
It is not as complete as the MicronTracker VB.Net demo, but it shows how to
perform the fundamental MicronTracker tasks, including
1. loading pre-existing marker templates,
2. interactively generating new templates,
3. recognizing markers, and

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 86

4. displaying positions and distances between markers.

The UI has been implemented using FLTK (www.fltk.org) a public domain
multi platform toolkit. The video and text display is done using OpenGL or
FLTK depending on configuration parameters. Note that the display of images
and overlays is just an example implementation where being cross-platform,
rather than top performance, was given the highest priority.
 MTDemosNet: Uses VB.Net and C# .Net to show how to use MicronTracker
functionality in Microsoft .Net environment. MTC is accessed via a wrapper
library (MTInterfaceDotNet.dll). The wrapper library is included, as a separate
reusable project, in source format.
The MTDemosNet VB.net demo functionality is described in section 8. Since
Vb.Net is relatively straightforward to read and understand, it is intended to serve
as a general-purpose presentation of the manner in which application
programmers should use the MicronTracker API in other environments as well.

 MTDemoVB6: Demonstrates a wide range of MicronTracker functionality

using Visual Basic 6 code. MTC is accessed via a wrapper ActiveX library
(Mtinterface.dll). The wrapper library is included, as a separate reusable project,
in source format. The ActiveX API is very similar to the one provided in earlier
product versions.

11.4 The SimpleDemoC Application

This application demonstrates the very basics steps required to use a
MicronTracker to detect and report positions. It is a Windows console application,
which loads pre-saved markers templates, initializes the camera and then loops for
a few seconds reporting the position of the recognized markers on the output text
window. It implements some of the basic steps described in the introduction to the
Programming model paragraph.
On initialization, the application
 Calls Cameras_AttachAvailableCameras to activate the cameras and
load their calibration files from disk; and
 Calls Markers_LoadTemplates to load all the marker template files.
To obtain a measurement, the application
 Calls Cameras_GrabFrame to obtain the most recent frame images from the
camera(s);
 Calls Markers_ProcessFrame to update the collection of Identified
markers.
 Obtains a collection of the handles of the identified markers by calling
Markers_IdentifiedMarkersGet;
 Accesses the pose of each identified marker by using
Marker_Marker2CameraXfGet to set the properties of a Xform3D object
accordingly, then prints the name, position and rotation of the marker using
properties of the marker and the Xform3D object.
On exit, the application calls Cameras_Detach to release the cameras and allow
the process to terminate.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 87

11.5 The MTDemoCPP Application

MicronTracker ships with a
C++ example application
that is available both in
executable and in source
format. By default the
application is installed
under Program
Files\ClaroNav\MicronTra
cker\MTDemoCPP. A
shortcut to this application
is placed on the desktop and
it is also available through
Start  Programs 
ClaroNav
MicronTracker. This
application is multi-platform and supports both
Windows and Linux. Its feature coverage is not as
complete as the VB6 demo, but it shows how to
perform the fundamental MicronTracker tasks:
o loading pre-existing marker templates,
o interactively generating new templates,
o recognizing markers,
o displaying positions and distances
between markers.
Aside from the main window where video images
and detected markers are displayed, it also has a
panel that can be used to record new marker
templates. The existing templates are listed in the
panel and can be deleted or renamed.
Of particular interest is that the application is
implemented using a C++ wrapper over the MTC
“native” API, making it easier to integrate into a
C++ development environment. The wrapper is
not complete, but it covers sufficient functionality
for most basic applications and can be easily
extended following the established patterns.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 88

12 Advanced Topics
12.1 Camera Exposure Controls
12.1.1 Exposure, Gain and Shutter
To enable detecting and correctly pinpointing an Xpoint in the image, a minimal
contrast of about 20 pixels values between its dark/bright regions is necessary.
Furthermore, its bright regions should not be over-exposed, i.e., the pixel values
there should fall below the maximum pixel value (255). The range of light
intensity encountered during operation is extremely wide (50-100,000 Lux),
making it necessary for the MicronTracker software to frequently adjusting the
camera exposure settings, which is represented by the Exposure property of the
Camera class.
The Exposure property value is actually a product of two other properties, GainF
(gain factor) and ShutterMsecs. ShutterMsecs corresponds to the length of the
period, in milliseconds, in which each pixel in the sensor is exposed to photons
before a measurement is taken (in the range 0.13 to 33 ms for the S60, 0.17 to 66
ms for the H40, 0.17 to 67 ms for the H60 model or 0.01 to 67 ms for H3-60 and
Sx60 on 15fps mode) 0.01-50 ms for Hx40 and Hx60. The GainF parameter
controls the amount of amplification the analog signal received from each pixel
undergoes before being converted into a digital value (in the range of 1-20 for
H40, 1-32 for S60, 1-24 for Sx60, Hx40 and Hx60 and 1-22 for H60). Gain is
similar to the ISO rating of a film. A higher gain increases the sensitivity of the
camera, but also increases the noise level. Noise level, and with it measurement
jitter, increases proportionally to the square root of the gain factor (e.g., a gain
factor of 4 will produce roughly twice as much jitter as a gain of 1).
In a given lighting situation, a range of different combinations of shutter and gain
can be used to achieve a similar distribution of gray levels in the image. Lower
gain will result in lower jitter. However, under low lighting conditions low gain
values will also result in longer shutter periods, in turn blurring moving objects
and causing potential loss of tracking during marker movement.

12.1.2 High Dynamic Range Mode (HDR)

High dynamic range mode (HDR) is a technique that allows solid detection in
situations of great illumination range. . This feature enables MicronTracker to
work reliably under the wide range of intensity levels found in the operating
rooms. For instance H3-60 model covers a wide range of 20Lux~400,000Lux
illumination.
In this mode MicronTracker assigns a cycle of frames to grab images at different
exposure levels that covers the whole range of illumination in the field of view.
Based on the range of illumination the cycle might have 1, 2 or 4 frames.
HDR mode might produce a latency of one full cycle in reporting the pose data,
although this is generally not significant in image guided application. The
additional latency might be noticeable if the camera is operating at lower frame

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 89

rate. In such circumstance enabling background processing will improve the

performance significantly and it will reduce the time lag.

12.1.3 HDR Exposure Locking Mode

Included in the C# Demo Application is an enhanced mode for 2 cycle HDR.
Enable it by selecting HDR Short Cycle Locked Markers, from the Camera tab of
the Settings menu. This mode locks an HDR exposure cycle for each of the named
markers, Marker #1 and Marker #2, allowing both markers to be detected across
wide ranges of illumination. Detection of other markers is not guaranteed, so this
mode will only be suitable when two markers are in the field.

12.1.4 Background Processing

This feature makes the MicronTracker grab and process frames automatically in a
background thread. Data is processes as fast as possible and the most recently
processed pose data is retrieved on user demand. This approach can result in up to
30% performance improvement in situation where there is significant application
processing.
In this mode the MicronTracker does not wait to receive a command to grab and
process frames. It simply grabs and processes frames as fast as possible and copies
the measured pose data to a separate buffer which is accessible in the application
level.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 90

12.1.5 Automatic vs. Manual Exposure Adjustment

When the AutoAdjustCameraExposure property of the Markers object is set to
true (its default value), MicronTracker employs a sophisticated algorithm in
adjusting the value of the Exposure property based on the contents of recently
captured frames. Alternatively, exposure can be adjusted by the application, which
may either adjust the Exposure property, or go to an even lower level and adjust
GainF or ShutterMsecs directly.

12.1.6 Shutter and Gain Limiters

When the Exposure property is set to a given value, its translation into specific
settings of each of its two components (shutter and gain) is determined by two
other Camera properties: ShutterMsecsLimiter and GainFLimiter.
ShutterMsecsLimiter. The following diagram shows how shutter and gain are set
for a given exposure value (along the blue path).

Since the shutter time controls the amount of motion blur for a given marker
movement, ShutterMsecsLimiter should be set to the highest value in which, under
typical operating conditions, tracking is still maintained. Similarly, GainFLimiter
should be set to a value corresponding to the maximum measurement jitter
acceptable.
When lighting is strong (over 500 Lux), the settings of the limiters should have no
noticeable effect on tracking performance. Their effect is more noticeable in
medium or dim lighting conditions, where trade-offs are required. Applications

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 91

that track slow-moving objects can reduce jitter by setting ShutterMsecsLimiter to

the maximum shutter time (33/66msecs) and GainFLimiter to some low value
(e.g., 2). Applications that track faster-moving objects should set the shutter
limiter at 20 msecs or lower, and the gain limiter to 5 or higher.
The VB6 demo application allows direct experimentation with the effects of
different exposure limiters’ settings through the Options dialog (see 8.9).

12.2 Overexposure Control Interleave

The automatic exposure adjustment algorithm optimizes settings for tracking
markers already detected in previous frames. When new, more strongly
illuminated, markers appear, they may be over-exposed (“whitewashed”),
resulting in their presence not being detected. To address this situation,
MicronTracker deploys a special overexposure detection algorithm to search for
regions in the images which may contain over-exposed XPs, and, when they are
found, to reduce exposure by an amount appropriate to allow their proper
detection and tracking.
By definition, overexposed markers do not have a reliable “signature” in the
image and they cannot, therefore, be detected with the level of reliability of
properly exposed markers. As a result, the overexposure detection algorithm may
occasionally cause unnecessary temporary dimming of the image, resulting in
some increase in measurement jitter and periodic oscillations in the brightness of
the image presented on the screen.
In most applications, detection of new markers appearing on the scene does not
have to be immediate. In such cases, it is acceptable to activate the overexposure
detection algorithm only periodically, e.g., once every 10 or 20 frames. The
frequency of overexposure detection activation is controlled by the
OverExposureControlInterleave property of the Markers object.
The VB6 demo application allows direct experimentation with the effects of
different OverExposureControlInterleave settings through the Options dialog (see
8.9).

12.3 Templates Matching Tolerance

Identifying a visible facet involves comparing the measured relative positions of
the ends of its vectors to the corresponding vector end positions of each facet of
each registered marker template. Identifying a facet, i.e., declaring a positive
“match” with a specific facet template, means that the distance between the
measured positions and the corresponding template positions is lower than some
tolerance. The value of that tolerance (in millimeters) is exposed to the application
software through the TemplateMatchToleranceMM property of the Markers class.
To guarantee that a facet will not be mistakenly identified as a different one, or
that multiple visible facets will not be matched to the same template, the
difference in length between the two vectors of a facet and between relative vector
end positions in different templates needs to be larger than twice the tolerance.
Increasing the tolerance reduces the likelihood that identification will fail due to
small position measurement errors or due to bending of the facet. However, it also
increases the required minimum length difference between vectors in a single

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 92

facet and the between the relative positions of Xpoints in different facets (see
10.5). It may be useful to increase that value from its factory default setting
(1mm) when measurements need to be performed on large markers at distances
larger than 1.5m from the camera (which is outside the calibrated volume!), or
when it is important to continue pose tracking despite some distortion of the
marker geometry (for example, when the marker is placed directly on a patient’s
skin). It may be useful to decrease its value when a large number of slightly
different markers need to be distinguished, and the measurements are performed at
distances shorter than the full depth of the calibrated volume. The Options dialog
of the demo application allows experimenting with the tolerance setting (see8.9)
The method Validate of the Markers class provides a quick and useful way to
detect potential misidentification problems in the set of marker templates given
the current TemplateMatchToleranceMM value. It should be executed whenever
changes to either the marker templates or the value of
TemplateMatchToleranceMM may have been made (see usage examples in the
VB6 demo application code).

12.4 Predictive Frames Interleave

In detecting the presence of facets in a frame, MicronTracker uses a combination
of two approaches:
 Comprehensive detection: The two image of the frame are comprehensively
processes to detect any facet’s “signature” appearance. Once a facet is detected, it
is compared with each facet of each known templates to locate a match.
 Predictive detection: The pose of each identified facet in the current frame is
predicted from past frames. A search in the current images is then done only in a
small area around each of the facet’s Xpoints’ predicted projection location in that
image.
The predictive technique requires fewer calculations to execute per frame,
reducing CPU load. It does not, however, allow detection of new facets, and it
does not ensure reliable tracking, especially of facets that are moving erratically.
The property PredictiveFramesInterleave of the Markers class allows the
application programmer to interleave comprehensive and predictive frames. A
value of 0 (no interleaving) will ensure that each frame is processed
comprehensively. This is the recommended setting when sufficient CPU power is
available or when measurements are done at a slow rate compared to the
movement of the objects it is measuring. Otherwise, a higher value should be set
to free up the CPU. For example, a value of 2 will place two predictive frames (P)
after each comprehensive frame ©: CPPCPPCPP…

12.5 Reducing CPU Load

In some circumstances, sharing the CPU with the MicronTracker may have an
undesirable effect on the capacity of the application to respond to external events
in a timely manner. Here, we list a number of strategies that may be used to
resolve this problem.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 93

12.5.1 Adjusting Predictive Frames Interleave

Setting the property PredictiveFramesInterleave of the Markers class (see 12.4) to
a value in the range 1-4 may produce satisfactory results, while saving as much as
50% of the CPU power.

12.5.2 On-demand measurement

The demonstration applications obtain measurements as fast as the camera and
available CPU power allows. In case your application cannot process the
measurements at the maximum rate, requesting a new measurement only when the
application is ready to process them would save CPU cycles otherwise taken up
by the MicronTracker.

12.5.3 Accessing the Image Buffers

The first call to Camera_ImagesGet or Camera_HalfSizeImagesGet after a
new frame was grabbed requires internal processing within the MicronTracker
software to prepare a properly formatted 8-bit image and to copy it to the
application’s buffers. If the images are not displayed on the screen, the application
should avoid making such calls.
If only individual pixels or a small region in the image needs to be accessed, calls
to Camera_PixelsGet are far more efficient than obtaining access to the full
images.

12.5.4 Using a dual-CPU computer

By using a host computer with more than one CPU enables the processing of
measurements to be accomplished with no effect on other asynchronous
application tasks. Note that MTC is single-threaded and will not occupy more than
one CPU to execute each function call.

12.5.5 Using a dedicated single-board computer

An alternative to a dual-CPU configuration is to use a separate SBC.

© Copyright ClaroNav, Inc., 2003-2016

 MicronTracker Developer’s Manual, MTC3.8 94

13 Standards
MicronTracker camera models Hx40, Hx60, Sx60 and H3-60 have been certified
to comply with the following Standards:
 CAN/CSA C22.2 No. 60601-1:2008-Class II device
 ANSI/AAMI ES 60601-1/A2:2010
 IEC 60601-1:2005- CB Test for Class II device
 IEC 60601-1-2:2004 – Class B device
 IEC 61000-4-2:2001 (Electrostatic Discharge)
 IEC 61000-4-3:2002 (Radiated field immunity)
 IEC 61000-4-4:2004 (Electrical Fast Transients / Bursts)
 IEC 61000-4-6:2003 (Conducted RF Immunity)
 IEC 61000-4-8:2001 (Power Frequency Magnetic Field)
 CISPR 11 (Radiated Emissions)
 IEC 60601-1:1988 + A1:1991 + A2:1995 and National Deviations to:
 CANADA – CAN/CSA-C22.2 No. 601.1-M90 (Update No.2, 11/2003)
 United States UL 60601-1: 2003

NOTE: that camera enclosure leakage needs to be measured in the end use application.

The MicronTracker also complies with Directive 2011/65/EU of the European Parliament and
of the Council of 8 June 2011 on the restriction of the use of certain hazardous substances in
electrical and electronic equipment.

© Copyright ClaroNav, Inc., 2003-2016