Manual Visiflow
Manual Visiflow
VISIFLOW
System User Manual
Copyright
AEA Technology
404 Harwell
Didcot
Oxfordshire
OX11 0RA
UK
No part of this manual may be reproduced or transmitted in any form or by any means,
electronic or mechanical, including photocopying and recording, for any purpose
without the express written permission of AEA Technology.
AEA Technology reserves the right to make changes to any hardware, software or
documentation without further notice.
Users finding errors or omissions in this manual are requested to contact AEA
Technology.
Acknowledgements
1. INTRODUCTION ..................................................................................................1-1
1.1 WHAT IS VISIFLOW? .............................................................................................1-2
1.2 VISIFLOW HARDWARE AND SOFTWARE INSTALLATION ............................................1-3
1.3 HOW TO USE VISIFLOW .........................................................................................1-6
1.4 HOW DOES PIV ANALYSIS WORK?............................................................................1-8
1.5 HOW DOES VISIFLOW WORK?..............................................................................1-14
2. HARDWARE .........................................................................................................2-1
2.1 HARDWARE REQUIRED.............................................................................................2-1
2.2 HARDWARE COMPONENTS ........................................................................................2-2
3. SOFTWARE COMMANDS OVERVIEW............................................................3-1
4. FILE MENU ...........................................................................................................4-1
5. EDIT MENU...........................................................................................................5-1
6. VIEW MENU .........................................................................................................6-1
7. PROJECT MENU ..................................................................................................7-1
8. BOUNDARY MENU..............................................................................................8-1
9. OPTIONS MENU...................................................................................................9-1
10. ANALYSE MENU..............................................................................................10-1
11. WINDOW MENU ..............................................................................................11-1
12. HELP MENU......................................................................................................12-1
13. GLOSSARY........................................................................................................13-1
14. APPENDIX A - GETTING STARTED WITH VISIFLOW .............................14-1
15. APPENDIX B - GETTING THE BEST FROM VISIFLOW ...........................15-1
16. APPENDIX C - ERROR MESSAGES...............................................................17-1
17. INDEX ................................................................................................................18-1
1. Introduction
Particle image velocimetry (PIV) is a non-intrusive optical technique, which measures
the complete and instantaneous velocity field throughout a selected two dimensional
slice of a flow. The method is particularly suitable for investigating unsteady flows,
whether periodic or transient.
In PIV, a laser beam is expanded into a sheet and is introduced into the flow. Video or
photography is used to record images of particles or droplets moving within the laser
sheet and if the laser beam is pulsed several times, each particle will yield multiple
images. Analysis of the multiple images yields velocity magnitudes and directions
throughout the illuminated flow plane. This analysis may be carried out rapidly and
fully automatically using a system like VISIFLOW.
The imaging basis of PIV makes it a valuable and powerful tool for visualising and
measuring both single- and two-phase flows whether in laboratory or industrial
environments.
Areas of application include sprays, jets, flames, engines, pumps, meters, furnaces,
separators, condensers and dryers, as well as flows around bodies. The two dimensional
VISIFLOW runs on PC based hardware, and can contain high speed image and array
processing sub-systems for rapid analysis of PIV images. A computer controlled
translation system, scanner system, video camera or video player can be used for fully
automated analysis of photographic and video images.
For Windows 3.1 and higher, VISIFLOW 16-bit should be used. For Windows 95, the
16-bit version can be used, or alternatively a 32-bit version is available. The 32-bit
version is required for Windows NT. This provides extra speed and improved multi-
tasking with other programs.
Lower priced versions of the software, VISIFLOW Acquire and VISIFLOW Display, are
also available. The first package has none of the PIV analysis or post-processing features
of the full priced version. It is useful to acquire one or more images from a video
camera, and save them to disk files. The second package has none of the image
acquisition and analysis features of the full priced version. It is useful to post-process
and display datasets which have been acquired on other computers using VISIFLOW or
other PIV analysis systems.
A list of the required hardware is given in the Hardware Components topic. The
hardware will normally be installed on purchase and then left intact. However, for
transportation or other reasons, it may be necessary for the user to dismantle/reassemble
the system. The installation process is described below.
Hardware
1. Computer - all processor cards should be left in the system permanently, but if it is
necessary to remove them, they should be replaced carefully, ensuring that a firm and
positive contact has been made with the mother board socket. The computer can be
opened by loosening the screws at the back of the system unit.
The frame grabber and array processor cards (if used) should be inserted into computer
bus slots. The parallel type security key should be plugged into a parallel (Centronics)
socket. The computer box should be securely closed. The mains power, mouse and
keyboard should be connected by their relevant cables.
Switch the computer on. Please note that if an EISA computer is being used and boards
have been inserted/removed, it is recommended to run the EISA configuration utility.
This informs the EISA controller of I/O addresses and interrupt allocation, and allows it
to warn of hardware conflicts. The configuration program may require following details
of the VISIFLOW boards:-
If Microsoft Windows has not been installed, do so now. This is explained in detail in
the Windows user manual.. Either Windows 3.1x, Windows 95 or Windows NT (32-bit
version of VISIFLOW only) can be used. The appearance of the software in Windows
95 and NT4.0 is different to the illustrations, because VISIFLOW fully conforms to the
new look of Windows 95.
If a Data Translation MACH series frame grabber is being used (DT3152, DT3155,
DT3157), the board should be installed in a PCI bus slot as indicated in the board
manual. The relevant device driver for Windows 3.1, 95 or NT should be installed. If
using Windows 95 or Windows NT, a DT driver V2.00 or greater is required. It is
recommended that a memory buffer of at least 2 MB is set aside for image storage.
If the Data Translation DT3851 frame grabber is being used, the DT software needs to
be installed. Follow the instructions in the DT3851 reference manual. Run the Windows
SETUP program to select the appropriate TIGA screen driver. Configure the graphics
driver using the DTSETUP program - select the frame grabber as the primary graphics
display. Load the TIGACD driver and the run the TIGAMODE program to select a 16
colour graphics mode. Add the TIGACD driver to the "autoexec.bat" file in the root
directory so that it automatically loads on start-up. Windows will now use the DT3851
frame grabber as the primary video adapter.
If the Matrox MAGIC frame grabber is being used, the MAGICLIB software needs to be
installed. Follow the instructions in the Matrox reference manual. Run the Windows
SETUP program to select the Matrox 1024 x 768 x 16 colour (small fonts) screen driver.
Ensure that the appropriate Matrox driver files are added to the "autoexec.bat" file in the
root directory so that they automatically load on start up. Windows will now use the
MAGIC frame grabber as the primary video adapter. The VISIFLOW software will
default to using the CCIR_MD.DCF camera format (CCIR 768x576) - this can be
changed if desired.
If the Imaging Technologies MFG frame grabber is being used, the board should be
configured for mixed mode single monitor image/VGA display. The MFG software
should be installed as described in the MFG installation guide. VISIFLOW is tested
compatible with version 1.4 of ITEX, using the 3 MB version of the MFG, with the AM-
VS video input card. The Tseng video driver should be installed for 1024 x 768 pixels,
16 colours resolution. VISIFLOW is configured to use the .CNF file pointed to by the
MFGCNF environment setting, e.g.
SET MFGCNF=C:\VISNPLUS\CONFIG\MFG.CNF
If the Digiplan IF1 stepper traverse controller is being used, the RS232 port should be
configured using the Change RS232 COM option from the Options menu. Settings of
2400 baud, 8 data bits, no parity and 1 stop bit should be used.
If a computer controlled video player is being used, the RS232 port should be configured
using the Change RS232 COM option from the Options menu. For the Sony LVR range,
settings of 9600 baud, 8 data bits, no parity and 1 stop bit should be used. The video
player should be adjusted to correspond with these settings.
If using Windows 3.x, run Windows by typing "win<CR>" at the MS-DOS command
prompt. Now insert the VISIFLOW disk in drive A. Select the "Run" option from the
Windows Program Manager "File" menu and then type "a:setup<CR>".
Alternatively, if using Windows 95 or NT, press the “Start” button, and select “Run…”.
Then type “a:setup<CR>”.
This will run the VISIFLOW setup program. The setup program will prompt for a
destination directory on the hard disk (the default is VISIFLOW). The setup program
will then copy automatically the relevant files and add the VISIFLOW icon to a
program group window.
If you are running Windows NT on a multi-processor machine, you can take advantage
of the additional processors through multi-threaded analysis, by adding the line,
MKL_PROCS=xx
to the environment variables, where xx is the number of processors to use.
The first time the software is run, a window will appear prompting for entry of the user's
company name and address. The program will then default to a set of reasonable values
for variables such as analysis resolution and time separation. When these values are
altered by the user, the program will remember the new values (they are stored in a
configuration file, VISIFLOW.INI), and use them next time.
PIV analysis is conceptually simple, but to obtain best results it requires a number of
parameters to be adjusted correctly. The VISIFLOW program has been designed to
make the entire analysis process easy and intuitive to use.
The PIV technique can be broken down into four main steps; the recording of the flow,
the pre-processing of the image, analysis of the image, and the inspection and post-
processing of the velocity dataset.
The first of these, the recording of the flow, is beyond the scope of this manual, (but see
the What is PIV? topic for a brief description). The flow image can be recorded as a
photographic negative or positive, a video tape recording, a live video signal, or a
digitised computer file.
The second step includes operations such as binarisation, and improves the quality of
the flow image.
For the third step, the VISIFLOW system currently relies on two main analysis
methods, known as two dimensional correlation and pairing/tracking.
If correlation is used, the analysis resolution then needs to be set. Options range from
16x16 to 512x512. These indicate the resolution used for the auto or cross correlation
function - the size of the interrogation area associated with each vector. Higher values
give higher resolution of particle position and hence higher accuracy of velocity vector,
but require more time to compute. The computational time for an NxN array goes as N2
log2 N.
Analysis can be initiated using the Analyse Flow option. A dialogue box is displayed
showing the current image segment being analysed and the computed velocity
magnitude and direction. The analysis can be cancelled at any time. A detailed analysis
may require an extended time and the VISIFLOW analysis window may be run as a
background process, allowing other programs or VISIFLOW windows to run
concurrently.
Using a video camera, the program will acquire and then analyse a single screen of
vectors. This is useful if the camera is pointing at a live flow, and immediate vector data
is required. However, the achievable vector resolution is lower than would be obtained
photographically, due to the poorer spatial resolution of current video sources. The live
analysis option can be adjusted to repeatedly analyse a flow, thereby recording temporal
fluctuations.
When the analysis is complete, the velocity dataset is saved to disk as a file. The third
step in the analysis is then carried out. This data file can then be opened, and the
velocity field displayed. Validation of poor data can be performed, and a number of
editing operations are possible. Various flow parameters such as vorticity can be
overlaid. The data can then be printed or plotted to a suitable output device, or saved in
a number of formats for transfer to other programs.
There are two main approaches to PIV analysis. These are correlation analysis, and
particle tracking analysis. The first analyses a region of the flow image, and determines
the average displacement of the particle field. The second analyses the positions of the
individual particles and then attempts to match them with their respective multiple
images. Each technique has advantages and disadvantages, and is relevant to particular
kinds of PIV flow image. Both of these approaches can be used to study single frame or
two-frame PIV images.
Correlation analysis
Correlation analysis involves correlating one region of the flow image against another
region. Movement of the flow region is revealed in its correlation. Correlation analysis
makes use of Fourier theory. This is outlined below.
∫ h( t ) e
2 πi f t
H( f ) ≡ dt
−∞
The complex function H(f) is the Fourier transform of h(t) and represents a
transformation of the time variable, t, to the frequency variable, f.
This is reversible; the original function can be retrieved by computing the inverse
transform:
∞
∫ H( f )e
− 2πi f t
h(t ) = df
−∞
The (cross-) correlation between two functions, g(t) and h(t), is defined;
∞
The correlation measures the degree of similarity between the two functions g and h.
The operation can be visualised as laying the two functions over each other, moving one
relative to the other and then checking for similarity. Where the two functions are
similar, the correlation has a high value.
Cor r( g, h) ⇔ G( f ) H∗ ( f )
Cor r( g, g) ⇔ G( f )
2
This auto-correlation can be thought of as “overlapping” the function with itself. If the
function has a periodic nature, then the auto-correlation will have a strong signal at the
same place as the periodic displacement. The auto-correlation of g(t) in effect compares
the function with its value at some other t=τ, g(t+τ).
Correlations can be extended to two (or more) dimensions in the same way as for the
Fourier transform. If the auto-correlation of a PIV image region is computed, then any
In summary then, analysis of a PIV frame can be carried out by breaking the image
down into regions and computing the auto-correlation function of each region. This
auto-correlation function is then searched for the strongest correlation signal. This
represents the dominant spatial displacement and the average particle velocity for the
image region can then be computed from this.
The auto-correlation function can be computed in a number of different ways. The most
obvious way is direct computation. When the function, h, is sampled at a number of
discrete values rather than being a continuous function, the Fourier transform becomes
the discrete Fourier transform (DFT):
N 2 πikn
Hn ≡ ∑ hk e N
k =0
l =0 k =0
This can be separated so that it can be performed by applying 1-D DFTs on the rows of
the data, followed by 1-D DFTs on the columns of the data. The DFT requires of the
order N2 operations, and is therefore increasingly computationally intensive as N
increases. The Fast Fourier Transform (FFT) is a computer algorithm for the efficient
computation of the DFT and reduces the computation to order N log 2 N operations.
Computation of the auto-correlation function for a region of a PIV image requires two,
two-dimensional Fourier transforms, and is therefore numerically intensive, requiring a
large-scale computation, even using the FFT algorithm.
A typical PIV image will yield a cross correlation function consisting of one or more
peaks, distributed non-symmetrically. As before the spatial position of the main peak
within the cross correlation image gives the average frequency of the multiple particle
image separation in the PIV image. This information is then combined with the time
separation to give the average velocity of the particles in the image. Cross correlation
offers the advantage of easy directional resolution as there is no 180 degree ambiguity as
for auto correlation, but requires a high recording rate to provide one image per laser
pulse.
Most correlation functions have more than one frequency peak within the image. These
peaks are due to particles correlating with other particles rather than themselves. These
peaks which are usually of lower amplitude can be thought of as noise, and can
occasionally give spurious results. It is important that the peak corresponding to the
particle correlating with itself is chosen by the computer. However, the positions of the
three highest peaks are calculated and stored, and can be substituted easily in the post-
processing stage.
The correlation analysis methods analyse rectangular regions of the PIV image to
identify average displacement of the particle field. This approach has a number of
advantages and disadvantages. The main advantage is that the method is robust when
the PIV image is noisy. If the noise is randomly distributed through the image plane,
such as film grain noise, the effect on the auto-correlation function is simply to raise the
background level, relative to the main signal peaks. As long as the main signal peaks
are visible above this noise background, then the analysis is unaffected.
Particle tracking analysis methods take a different approach to the correlation based
methods. They rely on identifying the actual positions of the particle images. This is
done by first thresholding the image and then using object analysis to identify particles
from groups of connected pixels. The centroids of these particle images can then be
computed, and the data formed into a list.
More advanced methods use this simple approach, but in addition apply various
parameters to the tracking process to increase the probability of differentiating particles
from the background, and in matching a particle image with its associated repeated
image.
Tracking methods can also yield the size (area) of the individual particle images. This
size information can be used to differentiate between different sized seeding particles.
As the LSI illumination is directed at the side of the particle, the brightness of the
scattered light is not directly linked to particle size, but brighter particles are generally
larger than darker ones, allowing a qualitative differentiation. However, most light
sheets do not have a “top-hat” intensity profile, and hence particles situated at different
depths through the sheet will scatter more or less light. This reduces the value of the
potential qualitative size information. Alternatively, if the particles are mono-disperse,
the apparent particle image size/brightness could possibly be used to determine the
position of the particle within the depth of the sheet, and thus determine three-
component velocities.
The behaviour of correlation and particle tracking methods is dependent on the seeding
density and quality of the flow PIV image. For low seeding densities, particle tracking is
preferable, with reliability improving as the density decreases. At higher densities, some
incorrect particle matching is likely and correlation starts to become attractive. At still
higher densities, where there are always several particle pairs in an interrogation region,
correlation is the best technique, due to its superior noise tolerance.
In practice, seeding density is dependent on the type of flow application. High speed
(transonic) gas flows usually have a very low seeding density, due to the difficulty of
introducing and dispersing sufficient material. Low speed liquid flows can be seeded
easily, allowing a high particle concentration to be established. For flows between these
extremes, the choice of analysis method is less clear, and usually depends on other
factors, such as the degree of noise present, or the complexity of the flow. Many tracking
VISIFLOW provides both correlation analysis and particle tracking analysis options.
Correlation implementation
The correlation analysis method first divides the image down into sub-images, or
segments. The size (N x N pixels) of these segments can be adjusted by the user from 16
x 16 pixels to 512 x 512 pixels. It is shown later that the error in measuring particle
displacement is roughly the same, when expressed as a fractional pixel error. Thus,
using higher resolution interrogation areas (effectively smaller pixels) results in higher
accuracy. This adjustable range of interrogation areas was used because below 16 x 16
pixels, (bias) errors in measuring particle displacement are high due to the large size of
the pixels. At 512 x 512 pixels and above, accuracy is sufficient, while computational
processing time continues to increase.
The spacing between the interrogation segments can be varied, but is usually set to give
an overlap of half of the segment size. This is x 2 oversampling. The square of the
magnitude of the resulting complex data array is then computed, giving real data once
again.
An inverse 2D FFT is computed on the results of the first computation, and the square
of its magnitude again computed. This is then the auto-correlation of the initial data
array. Alternatively, the above process is repeated for a second image, and the first FFT
result is multipled by the complex conjugate of the second. Now, taking an inverse FFT
gives the cross-correlation of the initial images.
The positions of the major signal peaks within the correlation array are then determined.
The highest peak is always at the centre of the 2D array, and represents the zero-order
or d.c. correlation spatial frequency. It is a measure of the image correlated with itself,
with no shift in position, and is proportional to the overall image intensity. This peak is
then removed from the array using a circular mask with an adjustable size. A simple
maximum value search is then performed on the array, and the co-ordinate position of
the highest peak is found. This gives the spatial frequency of the most dominant
correlation within the image, and hence the most likely value for the average velocity of
the image segment being analysed.
This correlation peak is then masked out as before, and the maximum value search
repeated twice more. The sizes of the deletion masks can be adjusted by the user, so that
the peaks are fully removed without also removing any other potential peak data. The
positions of the three highest correlation values are determined and recorded together
with their height (the degree of correlation) in the disk based analysis file.
If an image pixel associated with a particular object is adjacent to another pixel, then
that pixel is also included with the object. The object identification algorithm works by
tracing around the border of the object using the connectivity rule, and thus fully
identifying all of its boundary pixels. The extents of the object are now known. The area
of the object is computed, together with the co-ordinates of its centre of gravity, or
centroid. The centre of gravity, xg,yg for an object containing N pixels is computed
using:
1 N
xg = ∑ xi
N i =1
1 N
yg = ∑ yi
N i =1
Particles with an area below or above a certain user-specified value are deleted, as
representing noise in the image. Film grain noise tends to produce many single noise
pixels, while flare caused by unwanted reflections from the illuminating laser sheet
tends to cause large continuous areas of noise pixels (several hundreds or thousands of
pixels).
A simple nearest neighbour algorithm was initially used to match the appropriate
multiple particle images together. Although efficient, this was found to give poor results
when the seeding density was high enough so that nearest images did not always result
from the same particle. Thus, a method was developed to restrict the direction and
possible spacing searched that a particular particle image could be matched to. This
algorithm is described below.
The image is broken into a number of regions and a predicted velocity magnitude and
angle is then computed for each region. This prediction is done by analysing each
particle in turn and computing all possible associated velocity vectors by matching it
with all of its neighbours within a radius defined by a pre-defined maximum tracking
range. The angle and magnitude of each possible vector is then added to a histogram, a
separate one for each. This is illustrated below. The modal averages of the histograms
are then used to give the most probable values of the magnitude and angle of the average
velocity of the region.
Magnitude
Seed
Angle
Seed
The average velocity angle and magnitude and their standard deviations are then used to
improve (seed) the main matching process which follows. Each particle is matched to
2
Maximum tracking
Seed direction
Seed angle divergence
The matched particles with the smallest separation, and yet fulfilling the other
conditions are chosen to define the velocity vector. An option allows the user to specify
the maximum number of particles to search for during the match. If more than two laser
pulses are recorded during the camera exposure, this can be used to increase the
reliability of the matching process, as a genuine particle track has to extend over several
images. Once matched, the particle objects are removed from the centroid list.
This algorithm has been found to give good results with a variety of flow images. The
size of the prediction region needs to be adjusted for each flow image, so that it is large
enough that many particle images are present (to give representative statistics), but
small enough that the flow velocity is similar across the area. Clearly, if the flow
velocity has a high gradient and there are few particle images, then this approach is
unsuitable.
The matched velocity vectors differ from correlation derived vectors in that they have
random positions rather than being spaced on a regular grid. The correlation technique
is best when the flow image is noisy, usually as a result of unwanted reflections or film
grain noise. Noise in the spatial domain tends to be spread evenly in the frequency
domain, and thus not influence the dominant frequency component search. However, the
correlation technique requires several particle pairs inside each interrogation area, and
thus requires a uniformly high seeding density. Where the seeding rate is lower, particle
tracking is often preferable, as each particle pair produces a separate velocity vector.
The hardware required is dependent on the type of the image to be analysed. For basic
operation of the VISIFLOW system, the following is needed;
1) Intel i386DX based PC (or higher) - maths co-processor (387) is necessary - Pentium
or greater recommended.
2) VGA graphics card (or better).
3) Microsoft Windows supported mouse (or some other pointing device such as a
trackball).
4) AEA Security Key.
Component Supported
CCD camera Standard PAL/NTSC CCD cameras and
KODAK MEGAPLUS series
Frame Grabber Data Translation DT385X
Data Translation DT2855
Data Translation MACH series
Matrox MAGIC
Imaging Technologies MFG
Alacron FT200/DI
Array processor Alacron FT200
Data Translation DT2878-4
RS232 device Digiplan IF1/2 traverse controller
Spindler & Hoyer PCI traverse controller
Sony computer controlled video recorder
Jenoptic SLM controller
Film scanner TWAIN compatible
any scanner capable of producing .BMP or .TIF image files
It is recommended that the VISIFLOW program is exited and then restarted after the
hardware configuration is changed.
Array processor
The FT200 array processor board is similar to the DT2878, but uses a different
processor (i860). It has a number of variants, with performance up to 200 MFLOPS. It
also supports the DT-CONNECT bus.
The DT2878 array processor board consists of a 32 bit digital signal processor (DSP)
combined with 4 MB of fast memory. The DSP runs at 25 million floating point
operations per second (MFLOPS), and permits two dimensional fast Fourier transforms
and other computationally intensive calculations to be carried out rapidly.
Communications are made with the host PC using the AT bus. Image data can be passed
rapidly between the frame grabber and array processor using a high speed external bus
(DT-CONNECT).
The light intensity can be adjusted by varying the dc. voltage on its supply, and should
be set to give images with good contrast between particles and background, without
making the images too dark or bright.
The camera camera converts the optical image of the PIV flow into a black and white
video signal. The camera uses a charge coupled device (CCD) to convert the light
photons into an electrical signal.
For example, the Sony XC77RRCE camera has a resolution of 756 pixels horizontally
and 581 pixels vertically. The CCD array responds linearly to changes in intensity so it
is important not to overload it with too much light. The CCD is resistant to "blooming"
and is constructed to give an accurate 1:1 pixel aspect ratio with negligible field
curvature distortion. It is therefore ideal for PIV work.
The Kodak MEGAPLUS range of CCD cameras gives a higher resolution (currently
1024 x 1024 or 2048 x 2048 pixels), at the expense of a slower framing rate. Images
from these cameras can be acquired in an analogue or digital format.
The computer is used as the master controller for the entire system.
The computer optionally supplied with the VISIFLOW system is a PC/AT based system
running with an Intel processor (with built in maths co-processor). A large fast hard
disk stores the program and PIV data files. A large high resolution non-interlaced
colour monitor is used to display the velocity vector maps. The computer communicates
with the other hardware of the VISIFLOW system, using the built in host bus, high
speed dedicated data bus and serial ports. Floppy disks can be used to transfer data from
one computer to another.
The data on the computer hard disk should be backed up regularly to reduce the chances
of data loss. This data can be saved to the built in floppy drive, or alternatively to an
external device such as a tape streamer. A possible alternative is a network connection
and data storage on a remote file server.
Vector data can be printed out on a user supplied output device. Possible devices include
dot matrix printers, pen plotters, laser printers (including Postscript) and image
typesetters. The printer can be connected to either a serial or parallel computer port. The
device must have a relevant Windows device driver and be installed in the Windows
Printer list (see Windows user guide).
The frame-store holds the digitised video record of the PIV image.
For example, the DT3155 frame-store board consists of a large computer memory
together with a fast analogue to digital converter (ADC). The ADC digitises the
analogue video image and the data is stored in the RAM memory. One byte is used per
pixel, giving 256 grey levels of intensity information. Image data can be passed rapidly
between the frame grabber and array processor using a fast external bus.
In the optional negative interrogation system, this component holds the negative in
place, and prevents it from moving out of alignment during the analysis stage. By
placing the negative firmly against the straight bottom edge of the holder, it is possible
to ensure that it is aligned correctly vertically and horizontally. The two plates should
then be attached using the two screws. It is a good idea to keep the glass cover slides
clean to reduce noise in the optical image, and prevent artefacts.
The tension of the cover slides can be adjusted to accommodate a variety of thicknesses
of photographic image, including film and mounted slides.
The optional Digiplan negative translation system is a two axis electronic stepper
design.
The two translation stages are mounted perpendicularly in the XY configuration. Each
stage has 100 mm of travel with a resolution of 1.25 microns/step. Specified accuracy of
the stages is better than 10 microns across the travel distance, and repeatability is better
than 5 microns. Hall effect datum sensors provide a hard-wired origin facility for
reducing accumulated traverse error. The traverse velocity used is 10000 steps per
second.
The stepper motors are powered by current drives controlled by a master processor. This
processor is linked to the computer by a standard RS232 cable. The computer sends
data, such as required position to the processor, and the processor sends status data back
to the computer, such as current position.
The Spindler & Hoyer PCI negative translation system is also supported.
The optical rail in the optional negative interrogation system holds all of the optical
components and ensures good and stable alignment. It is a good idea to bolt the rail
down to increase stability.
The backlight is permanently attached to one end of the rail. The negative stepper stage
is attached to a carrier which is locked down onto the rail. The video camera and lens is
attached to another carrier.
The security key is a small device that attaches to the parallel port of the computer. The
VISIFLOW software regularly checks for its presence when it is running. The key is
used to ensure that the software is being run by an authorised user. The VISIFLOW
program will not work if this key is not present.
Instead of the negative interrogation system, using a two axis traverse and video camera,
a slide scanner can be used to input the photographic image. A scanner such as the
Nikon COOLSCAN can be used. This device uses stand-alone software to adjust
digitising parameters, and to produce .BMP images which can then be imported into
VISIFLOW and analysed normally.
A slide scanner is a preferred solution when using 35 mm film and when the digitising
resolution is sufficient (2700 dpi for the COOLSCAN), as it is compact, robust and
simple to use.
TWAIN scanners offer the option of acquiring images directly using the VISIFLOW
software. Non-TWAIN scanners require the image to be acquired separately and saved
as a disk file, using the software provided with the scanner..
Below is an alphabetically arranged list of all of the available menu based commands.
For more information, select the File menu command name from those named below;
This topic explains how to add text to a data file. This would typically be used to add
experimental information such as the flow conditions, photographic parameters, etc.
The information can be added before or after analysis.
Once selected, a dialogue box appears, prompting for text to be input. Up to 1000
characters of text can be added. The text entry box enables scrolling up, down, left and
right through the text, and also allows insertions and deletions.
Use this command to close all windows containing the active document. VISIFLOW
suggests that you save changes to your document before you close it. If you close a
document without saving, you lose all changes made since the last time you saved it.
Before closing an untitled document, VISIFLOW displays the Save As dialogue box and
suggests that you name and save the document.
You can also close a document by using the Close icon on the document's window
This topic explains how to leave the program and return to the Program Manager.
Once selected, the program will exit and the main window will be closed.
Use this command to open an existing document in a new window. You can open
multiple documents at once. Use the Window menu to switch among the multiple open
documents.
Shortcuts
Toolbar:
Keys: CTRL+O
The following dialogue options allow you to specify which file to open:
File Name
Type or select the filename you want to open. This box lists files with the extension
you select in the List Files of Type box.
Drives
Select the drive in which VISIFLOW stores the file that you want to open.
Directories
Select the directory in which VISIFLOW stores the file that you want to open.
Type
Select the type of file to open. VISIFLOW accepts binary format, and ASCII format
files (previously created with file export).
Network...
Choose this button to connect to a network location, assigning it a new drive letter.
Use this command to print a document. This command presents a Print dialog box,
where you may specify the number of copies, the destination printer, and other printer
setup options.
Shortcuts
Toolbar:
Keys: CTRL+P
The following dialogue options allow you to specify how the document should be
printed:
Printer
This is the active printer and printer connection. Choose the Setup option to change
the printer and printer connection.
Setup
Displays a Print Setup dialog box, so you can select a printer and printer
connection.
Copies
Specify the number of copies you want to print for the above page range.
Print Quality
Select the quality of the printing. Generally, lower quality printing takes less time to
produce.
A complex dataset may require lengthy printing on certain slow devices. However, the
Microsoft Windows print spooler is used, so other processing operations can be carried
out at the same time.
If problems with printing are evident, or a new printer is installed, it may be necessary
to change the printer set-up configuration. The Microsoft Windows user guide describes
how this is done.
Use this command to select a printer and a printer connection. This command presents a
Print Setup dialog box, where you specify the printer and its connection.
The following dialogue options allow you to select the destination printer and its
connection.
Printer
Select the printer you want to use. Choose the Default Printer; or choose the
Specific Printer option and select one of the current installed printers shown in the
box. You install printers and configure ports using the Windows Control Panel.
Orientation
Choose Portrait or Landscape.
Paper Size
Select the size of paper that the document is to be printed on.
Paper Source
Some printers offer multiple trays for different paper sources. Specify the tray here.
Options
Displays a dialog box where you can make additional choices about printing,
specific to the type of printer you have selected.
Network...
Choose this button to connect to a network location, assigning it a new drive letter.
This option selects one of the last four data files that were accessed, and displays the
dataset.
Once accessed, a filename will move to the top of the quick selection list. The option
will display less than four data files if less than four files have been accessed.
A file opening error will result if the data file no longer exists.
Use this command to save the active document to its current name and directory. When
you save a document for the first time, VISIFLOW displays the Save As dialogue box so
you can name your document. If you want to change the name and directory of an
existing document before you save it, choose the Save As command.
Shortcuts
Toolbar:
Keys: CTRL+S
Use this command to save and name the active document. VISIFLOW displays the Save
As dialogue box so you can name your document.
To save a document with its existing name and directory, use the Save command.
The following options allow you to specify the name and location of the file you're about
to save:
File Name
Type a new filename to save a document with a different name. A filename can
contain up to eight characters and an extension of up to three characters.
VISIFLOW adds the extension you specify in the Save File As Type box.
Drives
Select the drive in which you want to store the document.
Directories
Select the directory in which you want to store the document.
Network...
Choose this button to connect to a network location, assigning it a new drive letter.
This topic explains how to save the data file in one of a number of formats. When a
data file has been post-processed, it can be saved to the computer disk. The following
save formats are currently supported;
For further help on the menu options, select the appropriate Edit menu command from
those below;
This topic explains how to choose the next best data for a selection of velocity vectors.
The dataset holds three different values for velocity at each point. These values
correspond to the three most likely peaks in the auto or cross correlation function of the
original particle image. The velocity values are stored in descending order of
probability.
First, make a selection by clicking and dragging a rectangle around the desired vectors.
The 'next' edit menu option will then be made active.
Choosing next will toggle the next best choice of data in a cyclic fashion.
Shortcut
Toolbar:
This topic explains how to copy the vectors visible in the current view to the clipboard.
Choosing this option will copy the vectors visible on screen to the clipboard. The data
will be sent as a windows metafile, a text table and an OLE2 object to the clipboard,
enabling it to be pasted in a wide variety of programs, such as word processors and
spreadsheets. If the receiving application is OLE2 aware, in-place editing of the
VISIFLOW data file can be performed.
Graphical details of the picture will be preserved. The current flow scalar variable and
boundary overlays will also be copied if they have been previously enabled.
Shortcuts
Toolbar:
Keys: CTRL+C
First, make a selection by clicking and dragging a rectangle around the desired vectors.
The 'cut' option will then be made active.
Shortcuts
Toolbar:
Keys: CTRL+X
This topic explains how to flip a selection of velocity vectors by 180 degrees.
First, make a selection by clicking and dragging a rectangle around the desired vectors.
The 'flip' edit menu option will then be made active.
Choosing flip will rotate the selected vectors by 180 degrees. This is sometimes useful
when a flow recirculation is present and some small numbers of vectors are pointing the
"wrong" way.
Shortcut
Toolbar:
This topic explains how to insert a new OLE object into a dataset.
When this option is selected, a dialogue box appears, prompting for the user to select a
type of object to insert (embed), or cancel. If insert is chosen, an object of the selected
type is embedded and activated for editing. When editing is complete, the user can click
outside the area of the object to return to normal VISIFLOW operation. If later editing
of the object is required, it can be activated by double-clicking within its area.
This topic explains how to insert additional velocity vectors into a dataset.
When this option is selected, a dialogue box appears, prompting for the user to select
insert vectors, or cancel. If insert vectors is chosen, the cursor can be used to
interactively add extra vectors. This can be done by moving the cursor to the start of the
desired vector, depressing the left mouse button, and then dragging to the end of the
vector. When the mouse button is released, a velocity vector of the desired magnitude
and direction is drawn on the screen. This process can be repeated as required.
Shortcuts
Toolbar:
Keys: CTRL+V
This topic explains how to undo a previous deletion or mask of a selection of velocity
vectors.
The option is made active only if a selection was previously deleted or masked.
Choosing undo will un-delete or un-mask the selected vectors.
If the selection has changed since the delete or mask, it is possible to re-enable the
'undo' option by making the selection again and deleting it again. The 'undo' edit menu
option then becomes available.
First, make a selection by clicking and dragging a rectangle around the desired vectors
or left click on a single vector. The 'manual edit' option will then be made active.
Choosing the option will display a dialogue box, showing values for the velocity
magnitude and direction. It is possible to step through the vectors selected one-by-one.
Once the last vector is shown, the display steps back to the first vector in the selection. It
is possible to edit the displayed values by pressing the Change button after you have
modified the vector values. This then replaces the current velocity value with that
entered into the edit controls. If you do not press change, your modifications will be
ignored.
Shortcut
Mouse: Right button
This topic explains how to mask out a selection of velocity vectors. This option is often
useful when a physical object such as a nozzle or wall (or photographic blemish) is
present in the picture.
First, make a selection by clicking and dragging a rectangle around the desired vectors.
The 'mask' edit menu option will then be made active.
Choosing mask will flag the selected vectors, causing them to be blanked from the
display. Algorithms such as vector interpolation will now ignore the masked areas,
causing them to be left blank. It is possible to un-mask a region by selecting it, deleting
it, and then undeleting it.
This topic explains how to give a preferential flow direction to selected vectors.
This is useful when working with flows with a recirculation. It enables an entire section
of vectors to be given an overall flow direction - i.e. mainly up, down, left or right.
First, make a selection by clicking and dragging a rectangle around the desired vectors.
The ‘preferred direction’ edit menu option will then be made active.
Choosing the option will display a dialogue box, prompting the user for new values for
the flow direction.
For further help on the menu options, select the appropriate View menu command from
those below;
This command allows you to automatically update the display of a sequence of files that
have been loaded, so as to animate the data
This option becomes active when a sequence of data files have been loaded. When the
option is selected, a dialogue box is displayed, and the animation begins.
The time between animation updates can be changed by updating the value shown in the
dialogue box.
This topic explains how to overlay flow streamlines on the vector field.
Once selected (a tick mark is indicated on the menu), streamlines will be displayed as
one or more lines. The default behaviour is streamlines to be started at automatically
computed positions. These may not always be appropriate for a particular flow field. To
change these, please use Add Manual Streamlines and Streamline Options.
Streamlines are the paths taken by notional particles (with zero mass) introduced into
the flow velocity field. In the case of PIV data, these are instantaneous streamlines.
If the menu item is selected for a second time, the adjacent tick mark will be removed,
and the streamline overlay will be removed.
This option allows you to define your own start points for plotting streamlines. This is
useful when the positions of the automatic streamlines are inappropriate for the flow
field.
The option becomes active once Display Flow Streamlines has already been selected.
Once selected (a tick mark is indicated on the menu), the mouse cursor changes to:
. Now, every time the left mouse button is pressed, a streamline will be started
from this position.
By default Display Flow Streamlines shows both automatic and manual streamlines. If
using manual streamlines, it is likely that you will want to disable automatic
streamlines. This can be done by selecting Streamline Options from the Options menu.
If the menu item is selected for a second time, the cursor returns to normal, and
additional streamlines will no longer be added.
This topic explains how to display the currently selected scalar variable of the flow field.
Once selected (a tick mark is indicated on the menu), the variable of the flow will be
displayed either as a graduated colour scale, or colour coded contour map. Red colours
indicate negative values, while blue colours indicate positive values.
If the menu item is selected for a second time, the adjacent tick mark will be removed,
and the scalar variable overlay will be removed.
Shortcut
Toolbar:
This topic explains how to automatically fill any regions with missing data with
interpolated vectors. In addition, it can be used to convert a dataset to rectangular
gridding.
The fit radius is the number of vectors in a circle extending from the current vector
being filled. It must be at least the size of the largest data gap. Large values require
more lengthy processing as the number of vectors is greater, but give a better
interpolation.
The fit power is the weight assigned to vectors at different distances from the current
vector being filled.
The fitting procedure considers vectors within a user defined radius, and with a user
defined fit power against distance. The weighted average of surrounding vectors is
calculated according to;
where vi is the velocity of vector i, r is the distance of the vector i from the centre of the
averaging circle (where the fit radius is the circle radius), and n is the fit power. Higher
(more negative) values of n give more importance to vectors near the centre of the
circle. Setting n to zero, gives equal weighting to all vectors in the fitting circle.
The number of x and y co-ordinates in the new interpolated dataset can be specified
together with the minimum x and y co-ordinates. The latter is the position of the bottom
left vector. The distance between vectors can also be specified. These values are all in
microns.
If a boundary zero velocity condition has been set, this will be taken into account during
the interpolation. The interpolation routine replaces the masked boundary region with a
pseudo-grid of vectors of zero magnitude.
When OK is selected, the data will be interpolated and filled automatically. A large
dataset may require lengthy processing, and progress is shown as a percentage
completed in a dialogue box.
Shortcut
Toolbar:
This topic explains how to change the way the velocity vectors are scaled. This can be
useful where small scale fluctuations in a flow need to be seen.
Once selected, a dialogue box is displayed, showing the current scaling mode, either
automatic or manual. When set to automatic, the average velocity magnitude is used to
scale the vectors by a suitable amount for display. When set to manual, a scaling scale
value (in m/s) is required.
When scaling is changed, the calibration scale will also change accordingly.
By default, when an overlaid image is present (see Show Image As Overlay), vectors are
rescaled so that their length corresponds to that of their actual displacement referenced
to the image. This behaviour can be disabled by switching off the Re-scale During
Image Overlay option.
This topic explains how to read in a scanned image of the flow and overlay it onto the
analysed PIV dataset.
Once selected, a dialogue box is displayed, prompting for a filename to be entered. This
filename is the name of the file (stored in the TIFF image file format) that holds the
overlay image.
Once the filename is entered, the image is read in and overlaid onto the vector field. The
image is scaled so that it is positioned accurately with respect to the vectors. The image
and vector field can be zoomed in as normal.
The magnitude of the velocity vectors are scaled so that they match the particle
separations of the original image, enabling easy visual validation of the data.
Shortcut
Toolbar:
This topic explains how to automatically filter the velocity vectors, thereby smoothing
the data.
Once selected, a dialogue box is displayed. The filter kernel is displayed as a 5 x 5 table
with the current vector the centre value. The decimal value indicates the weight attached
to the neighbouring vectors, and can be altered manually to give different filtering
effects.
The default filter is an inverse exponential. When the default button is pressed, the
kernel values return to this default. Four sets of filter parameters are stored.
The filter kernel is displayed as a 5 x 5 table with the current vector the centre value.
The decimal value indicates the weight attached to the neighbouring vectors, and can be
altered manually to give different filtering effects.
This topic explains how to display the status of the velocity dataset. Once selected, a
dialogue box is displayed, showing the average velocity and angle, together with the
time the image analysis was started.
A velocity bias can be applied to the data. This subtracts a vector representing a constant
velocity value from each analysed vector. The values of velocity magnitude and angle to
be subtracted from each vector should be entered as image shift parameters.
If the Show Vectors option is disabled, then vectors are not shown. This can be useful
when you are using Show Image As Overlay, and you wish to just view or print the
image.
If the Display Ruler option is enabled, then a graphical representation of a ruler scale is
shown for the x and y direction.
This command allows you to select a particular data file from a sequence of files that
have been loaded. The mean flow field is also available.
This option becomes active when a sequence of data files have been loaded.
This topic explains how to automatically validate the velocity vectors, discarding any
poor data.
When this option is selected, a dialogue box is displayed, listing the adjustable
validation parameters; threshold and continuity.
A check can also be made of the vector continuity. This option compares the vector with
a weighted average (considering magnitude and angle) of the surrounding vectors, and
discards any which are different by more than a certain percentage The averaging
procedure considers vectors within a user defined radius, and with a user defined fit
power against distance. The weighted average of surrounding vectors is calculated
according to;
where vi is the velocity of vector i, r is the distance of the vector i from the centre of the
averaging circle, and n is the fit power. Higher (more negative) values of n give more
importance to vectors near the centre of the circle. Setting n to zero, gives equal
weighting to all vectors in the averaging circle.
A “quality factor” has been implemented, currently using the correlation peak ratio to
give a measure of the signal to noise ratio, and hence quality of the correlation result.
Lower and upper limits for this can be set.
If the “Substitute 2nd/3rd choice if fail” option is checked, then the validation algorithm
will try the second and third correlation peak choice if the first choice fails the
validation.
When the OK button is pressed, the validation parameters are saved for a later
validation calculation.
When the Apply button is pressed, the selected vector data will be validated using the
displayed parameters. If no vectors are selected, the entire dataset will be validated.
A large dataset may require lengthy processing. If a valid vector cannot be found at a
particular point, then the vector is temporarily deleted. Up to five passes through the
data will be performed, possibly restoring deleted vectors. At the end, the fill gaps
procedure is presented as an option if any gaps remain in the data.
Shortcut
Toolbar:
This command allows you to maximise the display area for optimal viewing.
When selected, it hides the VISIFLOW title bar, menu bar, toolbar, and status bar.
Note: to de-activate this command, use keyboard shortcuts to access this command (i.e.
Alt+V) from the view menu.
This topic explains how to zoom in to view a small region of the velocity dataset.
First, make a selection by clicking and dragging a rectangle around the desired vectors.
The 'zoom' view menu option will then be made active.
Choosing 'zoom in' will magnify the selected region to fill the main window. A new
option is then added to the menu, 'zoom out' which returns to the global view. Multiple
levels of zooming in are permitted.
Shortcut
Toolbar: and
Projects are a very useful feature that let you associate two or more PIV data files
together into a single package. Once a project has been defined, a number of extra
features become available. These include the ability to animate a sequence of datasets,
the mean velocity dataset, and the addition of several extra scalar variables - such as
turbulence intensity.
For more information, select the appropriate Project menu command from those below;
This command allows you to create a new project. When selected, a new project is
created and the dialogue box used for Edit Project is displayed, allowing you to add new
data files to the project.
Projects are useful ways to store information about a group of associated datasets, i.e.
datasets that were analysed from a sequence of consecutive flow images.
When a project is loaded, and has more than one associated dataset, sequence statistics
such as the mean velocity dataset become available, along with dataset animation.
This button on the toolbar becomes active once a project is loaded. It allows you to
quickly select and load a particular datafile that is part of the project.
This command allows you to open a project, previously stored on disk as a project file
(.PRJ). An open file dialogue box is displayed, allowing you to choose a project file.
This button on the toolbar becomes active once a project is loaded. It allows you to
quickly select and load a particular datafile that is part of the project.
This command allows you to add or remove data files from the current project.
A list box displays the current data files associated with the project.
It is possible to add single or multiple file names at one time. When added, the data file
name is checked so that it is unique within the project.
This command allows you to close the current project, and save any changes to it to disk
if desired. If there have been any changes to the project then it will be saved to disk. If a
project has not been named yet (i.e. it is newly created), then a save dialogue box is first
displayed, prompting you to enter a project name.
This command allows you to load other data files, recorded as a sequence. This then
allows additional data such as the mean velocity and Reynold's Stress to be displayed.
The command is highlighted only when a project is open and has more than one
associated data file.
For more information, select the appropriate Boundary menu command from those
below;
This topic explains how to define a boundary around a flow feature. This can be useful if
the flow is constrained within a particular geometry, or includes regions such as nozzles
where no flow can be present. The boundary data is stored as a number of segmented
line polygons, rectangles or ellipses. The vector data can be masked, depending on this
boundary data.
When this option is chosen (it is active when a data file is present), a modeless dialogue
box appears. This box can be moved to an unobtrusive part of the screen if necessary.
Three different types of boundary are supported - rectangles, elliptical and segmented
polygon.
Three different formats of boundary are supported, and can be selected using buttons;
1) Inclusion boundary. All vectors within the polygon defined by the boundary shape are
to be included in the final vector field. This polygon type is shaded green.
2) Exclusion boundary. All vectors within the polygon defined by the boundary shape
are to be excluded from the final vector field. This polygon type is shaded red.
3) The third format of boundary polygon is purely cosmetic and has no effect on the
masking of vectors.
button is pressed, the mouse cursor changes to: . The cursor should then be moved
to a corner of the boundary. When the left mouse button is clicked, a "rubber band"
appears, attached to the cursor.
For polygons, every time the mouse button is clicked, another polygon corner is defined.
The close shape button draws a line from the current vertex to the initial vertex, thereby
closing the polygon. If the shift key is held down during the boundary definition, then
the “rubber band” will be constrained to the horizontal or vertical direction only. This
allows accurately perpendicular lines to be drawn.
For rectangles or ellipses, the mouse cursor should be “dragged” to set the size of the
shape.
Further shapes can be added by repeating the process. Inclusion and exclusion
boundaries can exist simultaneously - this allows a flow area such as an annulus to be
defined.
When the OK button is pressed, the boundary data is retained and displayed as an
overlay.
Shortcut
Toolbar:
When this option is selected, a dialogue box appears, prompting the user for a boundary
file name (.BND suffix). When a suitable file has been chosen, the boundary data is
loaded, and the polygon data displayed as an overlay.
When this option is selected, a dialogue box appears, prompting the user for a boundary
file name (.BND suffix). When a suitable file has been chosen, the current boundary
data is saved to disk.
When this option is selected, a confirmation dialogue box appears, and then the current
boundary data is cleared from memory.
This option allows the user to add a new boundary outline to an existing boundary data
set. When selected, a dialogue appears (same as for new boundary file), prompting for
definition of the new shape.
This option allows the user to edit or delete existing boundary outlines.
When the option is chosen, “handles” appear around a boundary outline at its corners.
The mouse cursor changes to: . The handles can be dragged to a new position,
thereby changing the shape of the outline.
If delete is selected, the currently selected outline will be deleted (after a confirmation
prompt).
The selected outline can be changed by clicking inside the area of the desired outline.
The Edit Boundary mode can be exited by clicking inside the lefthand area of the client
area.
When this option is chosen, a dialogue box is displayed. The user can choose whether to
display the boundary data as an overlay or whether to colour shade the boundary
polygons.
The boundary flow condition option is used to define a constant velocity flow within the
boundary area. If selected, a pseudo-grid of constant magnitude vectors is generated,
filling the masked area. This grid is then taken into account during any subsequent
interpolation. The pseudo-grid velocity magnitude, angle and spacing value can be
entered. The spacing value is used as a divisor into the spacing of the main flow vector
grid. The higher the value of the divisor, the better the approximation to a continuous
boundary, at the expense of greater computation and memory consumption.
For further help on the menu options, select the appropriate Options menu command
from those below;
Analyse Flow Options Change the behaviour of the Analyse Flow command
Change Arrow type Change the velocity vector representation.
Change Colours Change the colour scheme.
Change RS232 COM port Change the RS232 COM port.
Change Peak Mask Change the sizes of the analyse peak masks.
Change Scalar Overlay Change the overlay flow variable and method of display.
Display Palette Change the brightness and/or contrast of displayed
images
File Locations Change the default directories for image and/or data
files.
Hardware configuration Change the analysis hardware.
Merge Data Files Add several PIV data files together.
Profiles Save or restore the current program settings.
Streamline Options Sets options on how to plot streamlines
Sum Sequence Add several TIFF image files together.
Timing Controller Options Set parameters for timing controller, such as image shift
User defined routines Configure for user supplied algorithms.
View Status bar Shows or hides the status bar
View Toolbar Shows or hides the toolbar
This option changes the behaviour of the Analyse Flow command (Analyse menu).
The parameters of the validation and interpolation (such as fit power and radius) should
be preset before commencing Analyse Flow.
If the Pause Before Analysis option is set, then VISIFLOW will wait for you to initiate
the analysis, after acquiring and displaying the image. This allows you to setup analysis
parameters such as any boundary shapes required.
This option changes the way that the velocity vectors appear on the screen and the
printed page. The vector is always plotted with its centre point in the centre of the
notional correlation rectangle, or at the midpoint of a tracked sequence.
Four options are available, bare lines, open and closed arrow heads, and filled arrow
heads. The different options take varying times to redraw on the screen with the bare
lines type being fastest and the filled arrow heads slowest. For a very detailed vector
dataset, the redraw time can be lessened by selecting the simple arrow shape for draft
display and editing until a final hard copy print out is required.
By default, the size of the arrowheads are scaled to the size of the arrow. This can make
it difficult to see the direction of small arrows. An alternative is to fix the arrowhead
size.
The selection of the chosen arrow type will persist until changed by the user.
This option enables the user to vary the colours displayed with the velocity vectors and
for the scalar variable overlay.
When the option is set to normal, the vectors will all be displayed in the system colour
(usually black on a white background). When the colour scheme is set to one of the
other options, the vectors will be displayed in one of three colours, green, blue, or red.
For the peak choice option, green indicates the vectors which correspond to the first
(highest) frequency peak in the correlation function, blue the second and red the third.
In addition, purple corresponds to vectors that have been interpolated.
For the peak ratio option, green indicates the vectors with the highest ratio of first to
second correlation frequency peak height, blue indicates a lower ratio, and red indicates
vectors with a ratio below this. The thresholds for these colour changes can be adjusted
by pressing the thresholds button.
For the magnitude option, green indicates the vectors with the highest velocity
magnitude, blue indicates lower magnitudes, while red indicates lower still. The
thresholds for these colour changes can be adjusted by pressing the thresholds button.
By default, scalar overlay variables are displayed using blue for positive values and red
for negative values. This can be changed if desired by pressing the +ve or -ve values
The selection of the chosen colour scheme will persist until changed by the user.
This dialogue box enables the user to vary the colours of the velocity vectors, when the
colour scheme has been set (by the Change Colours command) to show peak ratio.
Vectors coloured green indicates the vectors with the highest ratio of first to second
correlation frequency peak height, blue indicates a lower ratio, and red indicates vectors
with a ratio below this. The thresholds for these colour changes can be adjusted using
this dialogue box. Numbers greater or equal to one should be entered, with the green
value higher than the blue value.
The default values are 1.5 for green vectors, and 1.1 for blue vectors.
This dialogue box enables the user to vary the colours of the velocity vectors, when the
colour scheme has been set (by the Change Colours command) to show velocity vector
magnitude.
Vectors coloured green indicates the vectors with the highest ratio of first to second
correlation frequency peak height, blue indicates a lower ratio, and red indicates vectors
with a ratio below this. The thresholds for these colour changes can be adjusted using
this dialogue box. Numbers greater or equal to one should be entered, with the green
value higher than the blue value.
The default values are 1.5 for green vectors, and 1.1 for blue vectors.
The current settings of VISIFLOW can be saved to a profile file (.PRF). These can later
be reloaded. This is useful if you want to use several different pieces of hardware for
example.
You could set VISIFLOW up to use the first hardware configuration, save it to a profile,
then set up VISIFLOW to use the other configuration and save that to a profile. It is
then easy to switch between the configurations, simply by loading the relevant profile
file.
See also;
Load Profile
Save Profile
Restore Default Profile
The program asks you for the name of an existing profile file (.PRF). Once chosen, the
current settings are discarded and the saved settings are loaded and the program is re-
initialised.
The program asks you to provide a filename (.PRF file) to save the current settings into.
When this option is chosen, the current settings are discarded, default settings for each
adjustable parameter are loaded and the program is re-initialised.
Once selected, this option displays a dialogue box. A series of radio buttons enables a
COM port to be chosen. It is suggested that a different port is used for the printer or
plotter to avoid possible conflicts. COM3 or COM4 may require additional hardware
and interrupt drivers to be installed on most PC's.
The baud rate, parity and numbers of data and stop bits can be selected for the COM
port. These should be set to match those expected for the connected device. For example,
the Digiplan IF1 stepper controller is configured for 2400 baud, no parity, 8 data bits
and 1 stop bit.
The selection of the chosen port will persist until changed by the user.
When the auto or cross correlation function is produced, the program needs to identify
the peaks corresponding to the dominant frequency components (and hence to the
dominant velocities). The dominant peaks are chosen by searching through the array of
values making up the correlation function, and choosing the position with the highest
value. This peak is then blocked with a circular mask, and the search repeated for the
next highest peak. Three peaks positions are calculated and stored. For the auto
correlation function, the masking operation is also carried out on the central (or dc.)
peak.
Different flow pictures can record particle images with a variety of quality, that is,
different image sizes and levels of background noise. These parameters can affect the
size of the correlation dc. and frequency peaks, and it is sometimes necessary to adjust
the size of the masks used in the blocking operation. Correct mask sizes allow optimum
first, second and third choice of peak.
The change peak mask size option allows the mask size applied to each of the four peak
searches to be adjusted independently. The mask size is input in arbitrary units, and is
independent of the correlation resolution. The effects of different mask sizes can be seen
when analysing a flow if the display intermediate images option is selected.
The selection of the peak mask sizes will persist until changed by the user.
This option provides a dialogue box for the selection of a scalar flow variable. This
variable is computed for the data set and overlaid onto the vector display.
The flow variables that are currently supported are; vorticity and its standard deviation,
four components of the rate of strain tensor, velocity vector magnitude and its standard
deviation, u and v velocity component magnitude, correlation intensity, Reynold’s
stress, turbulence intensity, and number of vectors. The flow variable can be selected by
clicking on the combo box and then clicking on one of the displayed options.
ω = ∇ × v (1)
Two dimensional velocity data in a plane allows determination of only the out-of-plane
component of vorticity,
where (u,v) are the in-plane velocity components. This is evaluated by computing the
circulation around a closed contour surrounding the point,
Γ = ∫ u ⋅ dl (3)
and invoking Stokes' theorem to relate the circulation per unit area to ωz:
Γ
ω z = l i mA→ 0 (4)
A
The numerical approximation to (3) and (4) is implemented with a closed contour
defined by the eight points surrounding the node at which the vorticity is to be
evaluated.
Strain rate also plays a fundamental role in turbulent flows. The rate of strain tensor is,
1 ∂ ui ∂ u j
ei j = + (5)
2 ∂ x j ∂ xi
The three components e11, e22 and e12 of the strain rate tensor are evaluated from first-
order differences of the two-dimensional velocity at each point.
The remaining diagonal component e33 of the strain rate tensor is evaluated from the
continuity equation which yields,
1 ∂ρ
e33 = −
1
2
( e11 + e22 ) −
ρ ∂t
(6)
The velocity magnitude option simply maps the magnitude of the velocity to a colour
scale. The value of the magnitude is always positive.
The U comp and V comp options map the magnitude of the u and v velocity components
to a scalar scale.
The correlation intensity option maps the height of the currently selected correlation
peak to a scalar scale. Correlation intensity is always positive.
Where there is a sequence of PIV datasets loaded, three additional scalar variables are
available, together with the standard deviation of vorticity and vector magnitude.
Reynold’s stress arises from the correlation of the u and v velocity fluctuation
components. A non-zero value of this correlation implies that the two components are
not independent of each other. It is computed here as;
uv (7)
Turbulence intensity is the degree of turbulence of the flow over the measured time
period. It is computed here as;
(q ) ( )
12 12
2
= u2 + v 2 (8)
The number of vectors is simply the number of vectors contributing to the mean at each
particular point in the flow.
Two methods of visualising the variables are supported; colour coding, and contour
mapping.
Colour coding gives each scalar value a shade of colour on a graduated scale. The colour
becomes more intense as the variable increases in magnitude. Blue colours indicate
positive values of the scalar, while red colours indicate negative values. A calibration
scale is shown at the bottom of the flow display.
Contour mapping plots scalar equipotentials across the flow field. These equipotentials
are lines of constant value. The range of equipotentials is taken as the minimum to
maximum scalar values.
Best results are obtained with fairly smooth datasets. Noisy vectors will tend to give
erroneously high values.
When a flow image is displayed, a palette (look-up-table) is used to set how it appears
on screen. By varying the palette, the apparent brightness and/or contrast can be varied,
making it easier to view subtle features.
This palette can be changed from this dialogue box. A graph of output pixel intensity
against the image intensity is displayed. Thus, a 1:1 mapping (identity palette) is a
straight line with a gradient of 1.
The brightness and contrast scroll-bars can be moved, thus changing the palette.
Pressing the default button resets the palette to a 1:1 mapping.
If a graphics adapter with a hardware palette is present, the effect of varying the palette
in this way can viewed in real-time by previously displaying a suitable image.
When the OK button is pressed, the selected palette is remembered, and used when
displaying images in future.
NB - the image itself is not changed during this operation, only its display.
It is recommended that the VISIFLOW program is exited and then restarted if the
hardware configuration is changed.
This option enables a number of TIFF image files to be added together. This can be
useful when the particle image density is low, and the number of images per analysis
region needs to be increased. The images have to be in a black and white format.
A dialogue box is displayed, and a number of data files can be selected. The multiple
selection is accomplished by selecting the first file by clicking the mouse button, and
then subsequent files can be selected by holding down the Ctrl key and clicking the
mouse button. Alternatively the shift key can be held down while selecting, to highlight
an entire range of files.
The images are then added together, and another dialogue box is displayed, prompting
for a filename to save the summed TIFF file to.
This option enables a number of PIV data files to be added together. This can be useful
when you have results from two different areas of an image, and you wish to composite
the data.
Two or more data files should be selected. The multiple selection is accomplished by
selecting the first file by clicking the mouse button, and then subsequent files can be
selected by holding down the Ctrl key and clicking the mouse button. Alternatively the
shift key can be held down while selecting, to highlight an entire range of files.
The data files are then added together, and the composite result displayed. This can be
saved to a disk file if desired.
This option allows you to change the default directories where VISIFLOW looks for
image and/or data (.PIV) files.
The new directory path(s) should be entered into the appropriate edit controls. If the OK
button is pushed, VISIFLOW will check for the existance/validity of these directories,
prompting whether to create new ones where necessary.
Parameters for the display of streamlines can be changed from this dialogue box.
The default option is to display both automatically positioned streamlines, and manually
entered streamlines. Select the check boxes to enable/disable these options.
See also:
Display Flow Streamlines
Add Manual Streamlines
Parameters for the Timing Controller (if used) can be changed from this dialogue box.
The timing controller can be used to automatically synchronise the illuminating laser
and imaging camera to the flow being measured. Sub-parameters for the timing
controller can be set by pressing the Parameters button.
Image shifting parameters can also be set from this dialogue box. A velocity can be
defined that can be subtracted from each vector during analysis. In some hardware
configurations, the timing controller can be used to automatically control this image
shifting system. Parameters for mirror radius and mirror angular velocity allow easy
computation of the necessary angular speed to achieve a particular linear velocity shift.
This option enables the standard analysis routines to be extended, through the use of a
user supplied analysis algorithm. This algorithm is called from VISIFLOW and should
be in the form of a dynamic link library (DLL).
The name of the DLL should be entered into the text field. "None" should be entered if
there is no DLL.
Use this command to display and hide the Toolbar, which includes buttons for some of
the most common commands in VISIFLOW, such as File Open. A check mark appears
next to the menu item when the Toolbar is displayed.
The toolbar is displayed across the top of the application window, below the menu bar.
The toolbar provides quick mouse access to many tools used in VISIFLOW,
To hide or display the Toolbar, choose Toolbar from the View menu (ALT, V, T).
Click To
Save the active document or template with its current name. If you have
not named the document, VISIFLOW displays the Save As dialogue box.
Remove selected data from the document and stores it on the clipboard.
Use this command to display and hide the Status Bar, which describes the action to be
executed by the selected menu item or depressed toolbar button, and keyboard latch
state. A check mark appears next to the menu item when the Status Bar is displayed.
The status bar is displayed at the bottom of the VISIFLOW window. To display or hide
the status bar, use the Status Bar command in the View menu.
The left area of the status bar describes actions of menu items as you use the arrow keys
to navigate through menus. This area similarly shows messages that describe the actions
of toolbar buttons as you depress them, before releasing them. If after viewing the
description of the toolbar button command you wish not to execute the command, then
release the mouse button while the pointer is off the toolbar button.
The status bar is also used for displaying vector co-ordinate and velocity data, as you
move and click within the vector flow image area.
The right areas of the status bar indicate which of the following keys are latched down:
Indicator Description
NUM The Num Lock key is latched down.
For further help on the menu options, select the appropriate Analyse menu command
from those below;
Once selected, a dialogue box is displayed, enabling the user to change the method of
analysis used.
The most important option is the analysis method. There are four different methods
available; correlation, pairing/tracking, manual particle matching and user supplied
analysis.
Correlation analysis uses the auto or cross correlation function to give the average
velocity for a segment of image. It is the most robust technique, but also the slowest as it
is computationally intensive. The auto correlation can be calculated by VISIFLOW, or
computed optically using a optical auto-correlator. Auto-correlation is used for single
flow images, while cross-correlation can be used for single or pairs of flow images.
Particle pairing and tracking finds the centroids of each particle image and attempts to
match them to the corresponding multiple images. It is the quickest technique
(especially for sparse particle fields) but can be confused by turbulent velocity fields, or
where particle tracks cross each other. Particle pairing is used for single flow images,
while tracking is used for pairs of flow images.
Further options can be selected for correlation and particle pairing/tracking by pressing
the "Parameters" button.
The selection of the analysis method will persist until changed by the user.
See also:
Auto Correlation Options
Cross Correlation Options
Manual Match Data
Matching Options
Optical Correlation Options
Peak Mask Size
Set Flow Calibration
Set Image Source
This option displays a dialogue box prompting for entry of a number of parameters
applicable to analysis using auto correlation. It can be activated by choosing the
"Parameters" option in the Analysis Method dialogue box, when the auto correlation
option has been previously selected.
The analysis resolution (16 x 16 -> 512 x 512) determines the precision used to compute
the auto correlation function of the particle images. Higher values give greater accuracy
of velocity determination but require longer to compute.
The degree of overlap of the correlation area can be varied from 0% (areas are
bordering, but not overlapping), to 50%, to 75% (areas are overlapping by 75%).
Both the analysis area resolution and the degree of overlap will affect the physical
spacing of the vectors on the image. This spacing is indicated as a number of pixels and
also as a physical distance.
The display intermediate image option displays the auto correlation function for each
velocity vector that is computed. The auto correlation is displayed as a grey-scale bitmap
image in a moveable window on the screen. The peak mask is overlaid onto the chosen
peaks. The DC (or zero order) peak is shown masked with a dark grey border. The
frequency peak with the greatest height, that is the first choice of the search algorithm,
is masked with a bright white border. The other two peak masks chosen are shown with
progressively darker borders. The sizes of these peak masks can be individually adjusted
using the "peak mask size" option. The window size may be changed and maximised.
Selecting this option tends to slightly increase the analysis time as extra computational
overhead is involved. The display window can be closed using the system box in its top
left corner, as for other windows.
Two methods of optimisation are supported; Gaussian fitting, and centre of gravity
fitting. These fitting routines are used to interpolate between grid points (pixels) to find
the peak maximum with potentially better accuracy.
The Gaussian option uses a two dimensional Gaussian function, which is adjusted to fit
around the highest point of the peak. 25 grid values are used to fit the function.
The Centre of Gravity option thresholds the pixels around the peak maximum, and then
computes a centre of gravity for the thresholded shape. This centre is taken as the new
maximum.
The Gaussian option is potentially more accurate, but the Centre of Gravity option may
give better results when the correlation peaks have a distorted shape.
An optional windowing function can be applied to the interrogation area before analysis.
This can be selected by pressing the Windowing… button.
See also: Analysis Method
A flow image can be analysed either as a grey scale image, or as a binary black and
white image.
If an image is of good quality, i.e. dark background with little noise and bright particle
images (small in comparison with their separation distance) it is best to analyse as a
grey scale. Here the positional information encoded in the gradually faded particle edges
can improve the measurement of the position of the absolute centre of the particle, and
hence increase accuracy in the velocity analysis.
The average pixel level can be set to apply a check to each analysed segment. The image
is summed over its area and the total compared to the average pixel level. If the total is
below this value, the segment is assumed to be empty of particles, and analysis is not
performed. The area is masked out in the final data-set. If the pixel level is set to zero,
all segments will be analysed. This level only applies to binarised images. The selection
of the binarisation threshold will persist until changed by the user.
Adaptive thresholding can also be set. This overrides the manual threshold that has been
set and attempts to automatically set a threshold from a pre-analysis of the image. It uses
the minimum, maximum and mean intensity in an attempt to optimise the threshold,
and may give better results in some cases than setting the threshold manually. Also, for
correlation, the automatic threshold may change for each interrogation area, allowing
for a variation in image intensity across the whole image.
If the Intensity histograms option is set, then a graph of the pixel intensity distribution
within each correlation region is displayed during analysis. If binarisation is being used,
then the threshold level is shown with a red line.
This option displays a dialogue box prompting for entry of a number of parameters
applicable to analysis using cross correlation. It can be activated by choosing the
"Parameters" option in the Analysis Method dialogue box, when the cross correlation
option has been previously selected.
The analysis resolution (16 x 16 -> 256 x 256) determines the precision used to compute
the cross correlation function of the particle images. Higher values give greater
accuracy of velocity determination but require longer to compute.
The degree of overlap of the correlation area can be varied from 0% (areas are
bordering, but not overlapping), to 50%, to 75% (areas are overlapping by 75%).
If the Equal Sized Regions option is enabled, then the interrogation region size will be
the size in both the first image and the second. Otherwise, these can be varied
separately. However, the region size in the first image cannot be larger than that in the
second image.
Both the analysis area resolution and the degree of overlap will affect the physical
spacing of the vectors on the image. This spacing is indicated as a number of pixels and
also as a physical distance.
The display intermediate image option displays the cross correlation function for each
velocity vector that is computed. The cross correlation is displayed as a grey-scale
bitmap image in a moveable window on the screen. The peak mask is overlaid onto the
chosen peaks. The frequency peak with the greatest height, that is the first choice of the
search algorithm, is masked with a bright white border. The other two peak masks
chosen are shown with progressively darker borders. The sizes of these peak masks can
be individually adjusted using the "peak mask size" option. The window size may be
changed and maximised. Selecting this option tends to slightly increase the analysis
time as extra computational overhead is involved. The display window can be closed
using the system box in its top left corner, as for other windows.
Two methods of optimisation are supported; Gaussian fitting, and centre of gravity
fitting. These fitting routines are used to interpolate between grid points (pixels) to find
the peak maximum with potentially better accuracy.
The Gaussian option uses a two dimensional Gaussian function, which is adjusted to fit
around the highest point of the peak. 25 grid values are used to fit the function.
The Centre of Gravity option thresholds the pixels around the peak maximum, and then
computes a centre of gravity for the thresholded shape. This centre is taken as the new
maximum.
The Gaussian option is potentially more accurate, but the Centre of Gravity option may
give better results when the correlation peaks have a distorted shape.
An optional windowing function can be applied to the interrogation area before analysis.
This can be selected by pressing the Windowing… button.
This option will enable a live camera image to be displayed and continuously updated. A
scale is displayed giving the currently calibrated velocity magnitude.
With certain frame grabbers, it is possible to “zoom” the field of view, so that small
image details can be examined. It is also useful for fine adjustment of camera focus. The
position of the zoom region can be changed with the scroll bars.
If cancel is selected, the camera image will be frozen and the live display cleared.
The number of frames value indicates the number of frames to acquire and save to disk.
You can acquire up to the maximum number of frames indicated.
The acquire method refers to the way that images are acquired before being saved to
disk. If you select the Store To MemoryThen Disk option, then VISIFLOW uses the
image buffers provided by the frame grabber to provide an intermediate store for
incoming images. This allows high speed capture of images (up to consecutive frame
rate). After acquisition completes, these image buffers are saved to disk.
The frame separation value refers to the number of frames between acquisitions. A value
of 1 indicates consecutive frame acquisition. This value is only valid when storing to
memory then disk.
Buttons;
- Live
A live camera image can be displayed by pressing this button. Pressing it a second time
freezes the camera image.
- Record
When the Record button is pressed, the live camera image is frozen, and acquisition and
saving of the desired number of images begins. After completion, a message appears
confirming successful operation. The operation may be repeated as desired.
- Play
When the Play button is pressed, the previously recorded sequence is replayed, frame-
by-frame. The replay speed can be adjusted by entering the time in milliseconds between
frames in the Replay Speed edit box.
- Pause
When the Pause button is pressed during replay, the current frame is frozen. If this
button is pressed again, replay commences once again.
- Stop
When the Stop button is pressed, replay halts.
You can also enter a desired frame number directly into the Current Image edit box.
The manual particle matching option allows the user to manually specify the velocities
in the flow image. This option can sometimes be useful if the flow images are very poor,
and automatic analysis yields poor quality results. When this option is selected and an
analysis is performed, a dialogue box appears, inviting the user to move a cursor around
the screen and mark the positions of particles.
When analysing from a video image, either a live or frozen image can be overlaid
during the manual analysis. The choice of live or frozen can be made by pressing the
"Parameters" button in the Analysis Method dialogue box.
The cursor is simply adjusted by moving the mouse within the image overlay window.
The cursor changes to a "cross-hairs". When the cursor is lying on a suitable particle,
the left mouse button should be depressed and held down. This saves the current cursor
position. A "rubber-band" line then follows the cursor position. The cursor should then
be moved to the other particle in the matched pair, and the position marked again by
releasing the button.
If a frame grabber with a two monitor configuration is used (eg. the DT2855), the cursor
can be moved to the image screen by pressing the “Swap cursor” button. The cursor will
remain active on the image screen until a mouse click operation such as described in the
previous paragraph has been performed.
A vector is then displayed, linking the two images. It is important to mark the images in
the correct order if the direction is to be valid. The measurement is similar to the
measure calibrated scale process. The vector is then saved, and the cursor reappears,
enabling another particle pair to be recorded. Up to 500 particle pairs can be recorded
per video frame. When satisfied with the number of vectors recorded, the user clicks on
the OK button. If photographic analysis is being performed, the stepper system will then
move onto the next image segment, and the matching process can be repeated.
This option allows the user to record the length of a known calibration scale. The option
is called by the "Film Interrogation Parameters" and "Set Flow Calibration" options, and
is needed to calculate the negative magnification (macro factor), and image
magnification (microns per pixel).
A dialogue box is displayed, asking the user to ensure a suitable scale image is present.
This should be an image of a scale, present on the screen, with several divisions visible.
A scale is provided for this purpose. Typically, mm or cm scales are suitable. The image
can be set up using the "move stepper/video option" to centre the scale in the picture. If
the image is set up suitably, the user should respond "yes" to the dialogue question.
The cursor is simply adjusted by moving the mouse within the image overlay window.
The cursor changes to a "cross-hairs". When the cursor is lying on a suitable scale
graduation, the left mouse button should be depressed and held down. This saves the
current cursor position. A "rubber-band" line then follows the cursor position. The
cursor should then be moved to another suitable scale graduation and the position
marked again by releasing the button. The distance between the two marks should be a
significant fraction of the overall image size for high accuracy of measurement. A
measurement line will be left on the image.
If a frame grabber with a two monitor configuration is used (eg. the DT2855), the cursor
can be moved to the image screen by pressing the “Swap cursor” button. The cursor will
remain active on the image screen until a mouse click operation such as described in the
previous paragraph has been performed.
When both scale positions have been marked, the distance between them should be
entered in the relevant edit box. This distance is the "real" physical distance that the
scale represents, and should be entered in microns.
When the "OK" button is clicked, the appropriate calibration factor will be calculated
and stored.
If a film traverse system is installed, and a suitable RS232 controller driver is selected in
the Hardware Configuration dialogue box, this option allows the photographic film to be
moved under computer control. Alternatively, if a computer controlled video player is
installed, then this option commands the video player to move to particular frames
within a pre-recorded flow sequence.
The move option allows the user to move the flow negative in all four directions using
the two-axis stepper motor system. It is typically used to position the negative for
calibration or in the bottom left corner of the required area for commencement of
analysis. A live camera view is displayed, showing the effects of the movements.
The mouse is used to click on the relevant direction arrows. While the mouse button is
held down on the solid arrows, the negative stage will move at a speed of up to 2000
steps/second, with a ramped acceleration. When the mouse button is released, the XY
stage stops. This mode of operation is designed for coarse positioning. For finer control,
An indication of the distance travelled in the x and y direction is shown. The distance is
sometimes useful when assessing the area required for analysis.
A cross-hair is shown towards the bottom left of the image. This marks the bottom left
corner of the analysis area. When the analyse negative option is selected, the negative is
displaced so that the point under the cross hair is moved to the bottom left corner. After
analysis, the negative is reset to its original position.
If the "Hard Origin" button is pressed, the stepper drives will return to their fixed datum
points, as defined by the hardware wired hall effect sensors. The x and y distances are
then reset to zero. If the drives are now moved to a reference mark on the negative, and
the "Set Soft Origin" button is pressed, this position is remembered. It is then possible to
return to this point using the "Goto Soft Origin" button.
The move video option allows the user to move to any required region of the recorded
video sequence. The time code of the frame is displayed on-screen. Buttons are provided
for play, rewind, pause etc. In addition, the search option allows the user to enter a
required frame. When activated, the video then searches to find this frame and displays
it in pause mode.
This command allows you to analyse a flow image and then immediately display the
results on-screen. This is often useful as a shortcut when rapid results are needed. An
image is acquired, loaded from disk, or scanned with TWAIN, analysed, validated (if
desired) and then displayed.
Shortcut
Toolbar:
This option displays a dialogue box prompting the user to enter a number of parameters
applicable to the flow to be analysed.
The number of microns per pixel is an important conversion factor between the original
image, and images on the bit-mapped screen. It enables separation between vectors to be
calculated in microns, together with the absolute value of velocity magnitude in m/s.
This factor is computed using a calibrated scale, present in the original image.
The software supports both square and non-square image pixels. For the former, the
number of microns per pixel is the same in the x and y directions. For the latter, the
number of microns per pixel is independent for the x and y directions, and can be set
separately.
This option displays a dialogue box prompting the user to enter a number of parameters
applicable to the on-screen image. It can be activated by choosing the "Parameters"
option in the Image Source dialogue box, when the film analysis option has been
previously selected.
The Save acquired images to disk option saves the film image to disk as a TIFF file after
acquisition and before analysis. Each acquired segment image is merged together to give
the overall larger image.
The width and height of the area to be analysed can be entered. These quantities should
be in millimetres. They can be entered as minimum and maximum x and y values,
relative to the origin cross-hairs. For the Spindler and Hoyer PCI stage, these
coordinates are absolute referenced to the hard origin.
The flow macro factor is the ratio of physical distance to actual distance on the flow
negative. For example, if 5 cm of physical distance in the flow is represented by 1 cm on
the negative, the macro factor is 5. The factor can be directly entered or measured using
a screen cursor. It is often easy to photograph a graduated scale in the plane of focus of
the light sheet at the time of flow recording. This on-screen measurement requires the
number of pixels per micron to be previously calibrated.
This option displays a dialogue box prompting for the type of the image to be analysed.
The options are on-line (or live) analysis, film analysis using the interrogation stage,
analysis from a disk image file or analysis from a TWAIN source.
NB - The image of each film segment is not saved when using the Film imaged with
interrogation region option.
This option displays a dialogue box prompting the user to enter a number of parameters
applicable to analysing a disk based image. It can be activated by choosing the
"Parameters" option in the Image Source dialogue box, when the on-line analysis option
has been previously selected.
Batch processing works in the following way. VISIFLOW expects the image filenames
to use the format of a prefix followed by three digits, e.g. TEST000.TIF. The second
image in the batch would thus be called TEST001.TIF. Please use a prefix (e.g. TEST)
that is 5 characters or less in length, so that the total filename is less than or equal to 8
characters in length (MS-DOS restriction).
pairs
analysed
Using an increment of two (or more), discrete pairs of images are processed. This is
illustrated:
pairs
analysed
The analysis area of the disk image can also be defined, as a sub-region of the whole
image. This enables unwanted areas of the image, such as borders with text data, to be
excluded from the analysis.
If images are not frame aligned (such as on a film strip from a drum camera), then
reference marks can be used to align the frames during analysis.
If the Use Markers option is enabled, VISIFLOW will try to detect the location of two
reference marks (e.g. on the left and right hand sides) within the frame. It will then
offset the images using these reference marks, so that equivalent areas of the images are
being compared. This is particularly important when performing cross-frame analysis
such as cross-correlation or particle tracking.
This option displays a dialogue box prompting the user to enter a number of parameters
applicable to analysing a live image. It can be activated by choosing the "Parameters"
option in the Image Source dialogue box, when the on-line analysis option has been
previously selected.
If the number of samples is set to be greater than one, the system will analyse a sequence
of images, spaced in time. This enables temporal statistics such as the mean velocity
field and Reynolds stress to be obtained.
If the Analyse For Velocities option is enabled, then images are first acquired using the
frame grabber and then analysed. If the option is disabled, then images are acquired
only, without any subsequent analysis. This may be useful when it is desired that a
If the save files to disk after acquire option is enabled, the grabbed image(s) will be
saved to disk as a TIFF file after acquisition of all of the images has completed, and
before each analysis. Alternatively, if the save files to disk during acquire is enabled, the
grabbed images will be saved to disk after each of the images has been acquired. For a
sequence of images, an incrementing number will be appended to the filename,
beginning with zero (000). If a long sequence of images is acquired, the time required to
save all of the images as disk files may be considerable.
If the external acquire trigger option is enabled (certain frame grabbers only), an
external TTL signal can be used to begin the acquisition. This is useful for
synchronising with an external transient event.
If the computer controlled video player option is enabled (certain hardware versions
only), VISIFLOW can increment the video frame automatically, allowing a sequence of
images to be analysed without user intervention. The number of samples is here limited
only by available file space on the hard disk. The number of video frames between
samples should also be entered to set the frame increment when analysing a sequence of
images. The analysis will start at the currently paused frame.
If the Turnkey acquire option is enabled, then when VISIFLOW is first started, it
immediately acquires and analyses an image, without prompting the user. If in addition,
the VISIFLOW icon is placed in the Windows Startup group, and Windows is
configured to run automatically after power up, then the system becomes fully “turn-
key”.
If the Wait After Acquire option is enabled, then VISIFLOW will pause for a number of
seconds after acquiring an image, and before analysing it. This is sometimes useful
when you wish to ensure that no disk drive activity occurs in this period. During the
wait period, no indication of the delay process is shown, so if the system appears
unresponsive, it may still be waiting. If you are using Windows 3.1, then no other
system activity (such as other program mult-tasking) can take place during the pause
period. Windows 95 and NT use pre-emptive multi-tasking, which means other
programs may interrupt the pause period.
The time between samples specifies the time between acquisitions (in image frames) for
a sequence. The samples are acquired as a sequence of images and then saved to disk if
required. For the DT3851, the memory available limits acquisition to eight images. If a
frame separation of zero is entered, the system will wait for the user to respond before
acquiring the next image in the sequence. In this case, there is no limit to the number of
This option allows you to change the way that acquisition from a TWAIN compatible
device (such as a scanner) is performed.
The save image to disk option saves the acquired image as a TIFF file before analysis,
for possible later reference.
If you have more than one TWAIN device attached to your system, pressing the Select
TWAIN Source button allows you to change the currently active device.
This option displays a dialogue box prompting for entry of a number of parameters
applicable to analysis using particle matching. It can be activated by choosing the
"Parameters" option in the Analysis Method dialogue box, when the particle matching
option has been previously selected.
The dialogue box prompts the user for values, which if properly specified can improve
the reliability of the particle pairing/tracking process.
Direction coded by size:- this option specifies whether the image was recorded with an
intensity coding indicating which image was recorded first (tagging). If this option is
selected, the algorithm will assume that the biggest particle (i.e. the image with the
largest area) in a track is the first particle, and will thereby resolve the directional
ambiguity. The minimum ratio of the large and small areas of the main particle image
and the tag particle image can be specified. The algorithm always starts the track from a
large particle to a small particle.
Save matched list - this option specifies whether to save a list of the detected particle
image parameters. If selected, at the start of analysis, a text file, "matching.lst" is
created. Each particle is then listed in the file, including its x and y co-ordinates, and its
area. This file information can be useful when performing additional analysis with an
external program.
Use seed direction:- this option specifies whether the tracking algorithm starts to search
in a predefined direction when it encounters a particle image. If the main flow direction
is known, selecting this option can improve the reliability of the tracking. When
selected, the initial velocity angle option becomes active. If a seed direction is not
selected, a histogram algorithm will be used to automatically determine the seed velocity
direction and magnitude, using a statistical approach. For a given particle, every other
particle within the maximum tracking radius is matched with it, and histograms of the
possible velocity magnitudes and directions are determined. During the histogram
generation, particle matches with small separations are weighted higher than matches
with large separations.
Initial velocity angle:- this option gives the initial direction to search. The units are
degrees, measured from the horizontal, up to plus and minus 90 degrees.
Size of grid area:- when automatically generating the seed velocity, the image is broken
down into a number of areas, based on a grid. This option allows the size of the area to
be adjusted. It is best to make this grid area as large as possible so that a large number
of particle images will contribute to the statistical search. However, it should be small
enough that no significant velocity variation occurs within the area.
No. std. devs of avr. velocity:- If the seed velocity magnitude has been generated, the
tracking algorithm will use this magnitude to limit the tracking range. The limits of the
tracking range are defined by the seed average magnitude plus and minus a multiple of
the standard deviation of the seed magnitude. This multiple of the standard deviation
can be set with this option. Note - the maximum tracking range is still limited by the
maximum tracking velocity (see below).
The initial angle direction chosen will be displayed on the image as a number of arrows,
each being applicable to the particles in the surrounding area. Two arrows are displayed
for each area, corresponding to the minimum and maximum acceptance angles used by
the matching algorithm. If the use seed direction option is switched off, and the
algorithm thus uses the statistical histograms, the histograms are displayed adjacent to
the seed arrows. The magnitude histogram to the left, and the angle histogram to the
right. The modal average of each histogram is indicated by a small tick mark. If the size
of the grid area is set to less than 256 pixels, the display of the histograms can overlap
each other. However, this has no effect on the analysis of the data.
Minimum particle area:- this option gives a lower threshold for the acceptable particle
area, measured in pixels. Particles with an area below this will be rejected. Maximum
particle area:- this option gives a upper threshold for the acceptable particle area,
measured in pixels. Particles with an area above this will be rejected. These two
thresholds allow noise artefacts to be eliminated.
Maximum number of images per exposure:- this option gives an upper threshold to the
number of particle images per exposure. It can be computed knowing the pulse
repetition rate of the laser and the camera shutter exposure duration. It is always at least
two.
Minimum number of images per exposure:- this option gives a lower threshold for the
number of particles per exposure. It is always at least two, but can be raised for
increased reliability of track matching.
Maximum tracking velocity:- this option specifies the maximum separation between
particle images in a track. When looking for matches, the algorithm searches outwards
in a circular manner for neighbouring images. This value gives the maximum radius for
that search. It should be set as low as possible, but encompassing the highest velocity
likely to be present. The units are m/s.
NB - The maximum number of particles per image is 5000. If there are more particles
than this, then only the first 5000 will be considered.
This option displays a dialogue box prompting for entry of a number of parameters
applicable to analysis using optical auto correlation. It can be activated by choosing the
"Parameters" option in the Analysis Method dialogue box, when the optical auto
correlation option has been previously selected.
The analysis of a optical auto-correlation image consists of locating the dominant signal
peaks. The processing time is dependent on the time taken to download the image from
the frame grabber, and the time taken to search through the image data. VISIFLOW
determines the positions of both of the two dominant correlation peaks. The correlation
origin (zero frequency component) is then computed as the midpoint between the peaks.
This allows for the optical correlator to be set up slightly off-centre in the camera image.
The show masked peaks option displays the post-analysis masked out boundaries when
selected. This allows the best sizes for the masks to be determined.
The Peak Search Parameters button allows definition of an image sub-region. The
analysis time can be reduced by limiting the area of the image downloaded and then
searched for correlation peaks. This reduction of area is often possible when it is known
that a flow will always give peaks in the same area.
A computer controlled spatial light moduator (SLM) may be fitted. If it is desired for
VISIFLOW to automatically set this controller, then the option should be selected here.
The control voltage and frequency applied to the SLM can then be changed by selecting
the Configure SLM button.
This option displays a dialogue box prompting for entry of a number of parameters
applicable to analysis using optical auto correlation. It can be activated by choosing the
"Peak Search Parameters" option in the Optical Correlation Options dialogue box.
The analysis of a optical auto-correlation image consists of locating the dominant signal
peaks. The processing time is dependent on the time taken to download the image from
the frame grabber, and the time taken to search through the image data. This time can
be reduced by limiting the area of the image downloaded and then searched for
correlation peaks. This reduction of area is often possible when it is known that a flow
will always give peaks in the same area.
The x,y coordinates of the search area can be defined by entering numerical values in
the dialogue box, or by clicking and dragging a rectangle in the live image window. It is
useful here to have the optical correlator displaying a typical correlation pattern so that
This option displays a dialogue box prompting for entry of a number of parameters
applicable to computer control of the optical correlator spatial light modulator (SLM). It
can be activated by choosing the "Configure SLM" option in the Optical Correlation
Options dialogue box.
The voltage and frequency applied to the SLM can be entered. These values are only
applied to the SLM if the option to computer control the SLM is selected in the Optical
Correlation Options dialogue box.
New Window Creates a new window that views the same document.
Cascade Arranges windows in an overlapped fashion.
Tile Arranges windows in non-overlapped tiles.
Arrange Icons Arranges icons of closed windows.
Window 1, 2, ... Goes to specified window.
Use this command to open a new window with the same contents as the active window.
You can open multiple document windows to display different parts or views of a
document at the same time. If you change the contents in one window, all other
windows containing the same document reflect those changes. When you open a new
window, it becomes the active window and is displayed on top of all other open
windows.
Use this command to arrange the icons for minimised windows at the bottom of the
main window. If there is an open document window at the bottom of the main window,
then some or all of the icons may not be visible because they will be underneath this
document window.
VISIFLOW displays a list of currently open document windows at the bottom of the
Window menu. A check mark appears in front of the document name of the active
window. Choose a document from this list to make its window active.
For further help on the menu options, select the appropriate Help menu command from
those below;
Tip Of The Day Offers helpful hints and tips on using VISIFLOW.
About Help Dialogue box for version number and copyright information.
Commands Provides an alphabetical list of commands.
Glossary Provides a glossary for commonly used terms.
Hardware Describes the analysis hardware required.
Index Index of available Help topics.
Using Help Explains how to use the help system.
When this option is selected, a dialogue box is displayed, containing a useful hint or tip
for better use of VISIFLOW.
By default, the tip appears every time the program starts, but this can be disabled.
This option displays the version number of the software and copyright information.. The
free memory available and maths co-processor status are also displayed.
In addition if the VISIFLOW version is 32-bit, or the Display or Acquire variants, this
is also indicated.
The client area of the window is the main area of the window.
When no data file is present, the program title and copyright logo are displayed.
When a data file is present, the dataset is represented as a map of vectors, indicating the
velocity at each point in the flow. A scale can be shown to the left of the vectors,
showing average velocity magnitude and real-world distance. The vector origin (0, 0) is
shown as a small cross-hairs.
See also: Status of Dataset (for switching on/off the scale display)
PIV in its simplest form has an inherent problem with directional ambiguity. In a
multiple exposure image, it is not apparent which image comes first. The velocity vector
therefore has a 180 degree ambiguity in direction. This ambiguity can be removed by
encoding the PIV image at the time of recording. Several recording schemes exist
including track encoding.
The VISIFLOW package supports different laser pulse duration encoding through the
particle pairing analysis method.
Flow velocity analysis across two or more frames (using methods such as cross
correlation) does not suffer from this directional ambiguity.
This up arrow icon can be found in the top right corner of the window. When it is
clicked on, the window will increase to its maximum size and fill the screen. The icon
will then be replaced by the normalise icon ;
Clicking on the normalise icon will restore the window to its previous size.
This down arrow icon can be found in the top right corner of the window. When it is
clicked on, the window will reduce to its minimum size and become iconised. The
VISIFLOW icon will be displayed as it's representation (red arrows together with black
particles on a white background).
Double clicking on the minimised icon will restore the window to its previous size.
PIV (Particle Image Velocimetry) is a technique which can produce whole-field two
component velocity data for a plane within a flow. It relies on matching multiple images
of seed particles moving within the flow.
The sizing border is the thick edge of the window, and is present whenever the window
is not fully maximised or iconised (minimised).
If the left mouse button is now depressed and held down, the window can be dragged to
a new size. When the button is released, the window takes up the new size, and the
screen is usually redrawn with the view scaled correctly.
The window size can also be adjusted with the size command from the system menu.
When this icon (located at the top left of the window) is clicked, a system menu will be
displayed. It contains options for closing the window, moving and sizing the window. It
is also possible to maximise or minimise the window from this menu.
This region of the window describes the title of the program together with the associated
data file. If no data file is present, "new data file" is listed, otherwise when a file is
opened, its name is added to the title bar.
As for other Microsoft Windows programs, when a window is active, the title bar
becomes highlighted (white text on blue for the standard Windows colours).
Vector selection is needed when the user wish to edit or modify one or more velocity
vectors in a dataset. It is also needed to select a region to zoom in to for a more detailed
view.
A rectangle of vectors can be selected by placing the cursor just to the top and left of the
top left vector and then pressing the left mouse button. If the button is held down and
moved down and to the right, a rectangle will be "dragged" around the vectors. When
all of the required vectors are enclosed the mouse button can be released. This will fix
the selection rectangle in place. If the button is released without being dragged, a single
vector will be selected, nearest to the cursor position.
Another region can be selected by simply repeating the above process. To cancel the
vector select, simply press and release the right mouse button.
When the selected region is zoomed, the selection remains active, and can be seen to
enclose all of the vectors in the zoomed display. It is possible to reselect different vectors
within the zoomed display (and thereby edit or zoom further in).
If image overlay is enabled when a single vector is selected, editing "handles" appear at
each end of the vector. If these handles (small black squares) are dragged with the
cursor, the vector can be changed in magnitude and angle. This feature allows easy
correction of visually obvious vector errors through comparison with the image overlay.
All velocities are displayed in polar form, i.e. with a magnitude in metres per second,
and an angle in degrees (radians in the ASCII text output file).
+90
θ
+180
0
−180
−90
Figure 13-1. VISIFLOW angle measurement
Velocity data is stored in a disk file in either binary or ASCII formats. These formats are
subject to change without notice.
ASCII format
ASCII format is used when you copy data to the clipboard, or export the data to an
ASCII file. It takes up more disk space, and is slower to read from and save to, but is
easily readable.
The format described is applicable to the current version of the software (5.0) and
consists of two parts. Each line consists of ASCII characters, terminated by a carriage
return, line feed pair.
The first part consists of a data header, containing various information about the datafile
as a whole. Each data value is described, followed by a tab character. The data value
itself then follows.
The second part of the file then contains the actual velocity vector values. Each vector
takes up one line, with tab characters separating the numerical values.
The first two values are the position of the vector within the flow field in microns. This
is stored as an integer value.
Then comes pairs of values representing the velocity. The first in the pair is the
magnitude in m/s, the second the angle in radians. The number of velocity pairs is set by
the value of the “Number of Peaks” in the data header. These are stored as floating point
values.
If the “Flow variable” setting is other than “none”, another value is present at the end of
the line, containing the indicated scalar flow variable for that vector. This is stored as a
floating point value.
Binary format
The format described is applicable to the current version of the VISIFLOW software
(6.0) and consists of five parts.
The first part of the data file is a block of 17 bytes, currently with the ASCII code
"PIV_PLOT_FILE_6.0". This identifies the version of the VISIFLOW software that
created the data file, and is tested for compatibility when the file is first loaded.
The second part of the data file is a block of 6 bytes, containing offset information for
the rest of the parts. The format is as follows;
2 bytes: integer giving the offset of the header block from the start of the file.
2 bytes: integer giving the offset of the experimental text block from the start of
the file.
2 bytes: integer giving the offset of the main velocity data block from the start of
the file.
The fourth part of the data file consists of a block of 1000 bytes. This consists of any
experimental descriptive text stored with the data. The text is stored sequentially with a
single terminating NUL character (0 hex).
The final part consists of the velocity data itself. This is stored in blocks of 64 bytes per
vector. The format of the block is as follows;
VISIFLOW software runs under the Microsoft Windows environment, providing the
user with a powerful, consistent and user friendly graphical interface for the user, and
providing extensive data export features and simplified peripheral interfacing.
Windows, icons, drop-down menus and mouse allow ready control of the powerful
processing algorithms. Multi-tasking permits simultaneous analysis, post-processing,
printing, and use of other Windows or DOS programs.
14.1 Installation
The VISIFLOW package includes the manual you are reading, the software on a floppy
diskette, and a security key. The installation process is detailed below.
1. First fit the security key. The software tests for the presence of this key before it will
run correctly. The parallel port key is a small blue module with male and female
connections. First, identify the side of the key with the arrows marked “Computer”. This
side of the key should be inserted into a parallel port socket on the back of the computer.
If also using a printer on this socket, it can be connected to the other side of the key for
“passthrough” operation.
2. Now switch the computer on, and allow MS-DOS to boot up. If optional hardware
such as frame grabbers, array processors, scanners or traverse controllers are fitted,
these should be configured according to their respective installation instructions. If
Microsoft Windows is not installed, please do so, following the Microsoft Windows User
Guide. Once installed, it should be run by typing "win<CR>" at the MS-DOS prompt.
Windows will now load.
3. The VISIFLOW software should now be installed. This is done by inserting the
software diskette into the floppy drive of the computer. Select the "Run" option from the
14.2 Configuration
If you are new to Windows, it is suggested that you run the Windows Tutorial. For
Windows 3.1, this can be accessed by selecting the Windows Tutorial option under the
Program Manager Help Menu. This then takes you through the various features and
teaches how best to interact with Windows programs.
The VISIFLOW software can be run by double-clicking on the VISIFLOW icon:
The first time the VISIFLOW software is run, a dialogue box as shown in Figure 14-1
will appear prompting for entry of your company name and address. This is the User
Details dialogue box and is used to customise the program so that the user's details
appear briefly each time the program is run.
When you have completed entering the details, click on the OK button to continue. The
VISIFLOW program now loads and runs. The sign-on screen is displayed briefly, see
Figure 14-2.
The sign-on screen is then removed, and the main VISIFLOW screen displayed, as
shown in Figure 14-3.
The VISIFLOW program now has to be set up for the particular hardware configuration
that you have. This is done by selecting the Hardware command from the Options
menu. The Hardware Configuration dialogue box is then displayed, Figure 14-4.
The options corresponding to your own hardware should be selected. Further options
available for each hardware device can be accessed when available through the
appropriate Options button. When you have completed the hardware configuration, click
on the OK button to finish the operation. It is recommended that you now quit the
VISIFLOW has a full on-line help system, so that you can get easy access to the
documentation relating to any feature of the program. The help system is context-
sensitive so that you get help relating to the feature you are currently using. The help
system can be accessed through the Help menu. Choose the Index option to load the
Windows Help system, and display a list of help contents. The Help system is shown in
Figure 14-5. Alternatively, press the F1 key. This is a "hot-key" shortcut, and will
display the most relevant help information.
The VISIFLOW program can be broken down into two main areas: image analysis, and
post-processing.
Firstly, the type of image to analyse should be selected. This is done by selecting Disk
image as the source from the Source of Flow Image dialogue box, shown in Figure 14-6.
This can be accessed by selecting the Image Source option under the Analyse menu.
The magnification of the image should be entered, so that the program knows how to
scale the displacement it measures in pixels, into the magnitude of the velocity in m/s.
The number of microns per pixel corresponds to the view that the CCD video camera
The pulse separation is the third of the calibration parameters. This defines the time
period between the laser pulses forming the multiple exposed PIV image. For the test
analysis, this should be entered as 1000 µs.
Next, chose Particle Pairing as the method of image analysis. This is done by selecting
the Analysis Method option under the Analyse menu. Particle pairing can then be
selected from the dialogue box, shown in Figure 14-8.
Further analysis parameters should be set by clicking on the Parameters button. This
causes another dialogue box to appear, Figure 14-9.
All parameters should be set the same as for the illustrated dialogue box. The OK
buttons for both of these dialogue buttons should then be selected. Analysis can then be
performed by selecting the Analyse Flow option from the Analyse menu.
A dialogue box appears, prompting for the TIFF image file to analyse. This should be
entered as "image.tif", as shown in Figure 14-10.
When the OK button is clicked, analysis begins. The TIFF file is loaded, and displayed
on the screen. The pairing analysis first computes particle centroids, then computes and
overlays direction and magnitude histograms and then matches the particle images.
Finally the data is saved to disk. A typical screen display is shown in Figure 14-11.
When you have analysed a flow image, the velocity data can be viewed as a vector map.
This can be done by opening a data file previously created. The description below
applies to "image.piv", the pre-saved data file included with your package. This data file
was produced by analysing "image.tif", and can be compared against your own analysed
data file, "try1.piv" from the previous section.
To open the data file, select Open... from the File menu.
A dialogue box is displayed, prompting for a file name, as shown in Figure 14-12. Type
in the name of the data file, and click on the OK button. The file is then loaded and
displayed on the screen, Figure 14-13.
A region of the vector field can be selected by holding down the left mouse button and
"dragging" a rectangle around the data. When the button is released, the defined vector
region is bounded by a rectangle, showing that the vectors are selected.
The selected region can be unselected by clicking within the left hand region of the
screen.
This view can be moved around the whole vector field using the scroll arrows in the
corners of the window. The view can be zoomed out to the whole data set using the
Zoom Out option under the View menu.
The dataset can be printed to paper using the Print... command under the File menu.
This brings up a dialogue box as shown in Figure 14-15.
Many other operations are possible on the saved data, including automatic and/or
manual validation, interpolation to a regular grid, filtering, display of various flow
variables such as vorticity and strain and animation of sequences of datafiles. Always
refer to the on-line help (F1 key) for additional information.
15.1 Introduction
The purpose of this section is to offer the user advice on how best to use VISIFLOW in
order to obtain the best quality results from the user's own data. Indications are given as
to the performance that one can realistically expect to achieve taking into account the
quality of the data to be analysed..
First, it is important to understand that the ability of VISIFLOW to extract good quality
results depends to a great extent on the quality of data presented to it. The best results
will be obtained by appropriate application of the available algorithms.
It is important to recognise what constitutes good quality in PIV data. In general this
means:
15.3 Pre-processing
Once data has been taken and presented to VISIFLOW, it is important to choose the
analysis method which will get the best results from that data. Whichever method is
chosen, pre-processing the data will generally improve the performance. If the particle
images are dark on a light background (i.e. negative) then improvements will be gained
by inverting the image. Following this, "binarisation" is usually recommended to
improve the image signal to noise ratio and improve contrast. Problems due to image
artefacts can often be avoided by setting the minimum and maximum particle size
options when using the particle pairing/tracking algorithm.
When it comes to analysing data with VISIFLOW, there are a number of choices to be
made. Firstly, the analysis method - either correlation based analysis or particle
pairing/tracking. In general one should choose correlation whenever:
For best results using correlation, the magnification and/or size of the interrogation
region (resolution) should be adjusted to ensure several particle pairs are present. The
peak searching algorithm then searches the computed correlation function for the
highest peak. A peak mask is used to delete the peak once it has been located. This
allows the searching algorithm to search for the next highest peak. The dc peak mask
size should be set to fully remove the dc peak for auto correlation. The other peak masks
should also be adjusted so that they fully remove the correlation peaks after detection,
but without removing too much of the rest of the correlation data. If the peak masks are
set correctly, the peak searching algorithm will then be able to correctly find the best
three peaks. Resolution should be set to obtain the required speed/accuracy and to
change the number of vectors analysed.
The maximum tracking velocity should be set to just larger than the maximum particle
separation. If the approximate flow direction is known then select the set flow direction
option and specify the direction. Otherwise, do not select the set flow direction option
because this will activate the automatic histogram generation. The histogram grid size
should be set large enough to enclose a significant number of particle pairs, thereby
giving good statistics, but small enough that no significant velocity differences are
present.
When you wish to analyse more than one image (or pair of images), you should use
batch analysis. Here, you set VISIFLOW to analyse a sequence of flow images,
generating multiple .PIV datasets. It is also possible to set VISIFLOW to perform
certain post-processing operations automatically during batch analysis - see Analyse
Flow Options.
Choosing the most appropriate post processing routines can mean the difference
between good results and poor results.
If the Save acquired images option was used during analysis, the TIFF image can be
overlaid on the vector dataset, using the Show Image As Overlay option. The computed
velocity vectors are then clearly visible against the background image and the
effectiveness of the analysis method can then be judged. Any invalid vectors are also
easily observed.
For automatic vector validation, it is recommended that upper and lower magnitude
thresholds are first set to eliminate vectors which are clearly too large or small when
compared to known flow information.
It must be noted that the accuracy of the flow recorded image itself and not the analysis
system is usually the limiting step for the PIV technique. Care must be taken to
minimise image distortions by using high quality camera optics, which give flat image
fields at the close working distances usually employed. The camera should be set
perpendicular to the light sheet and carefully focused so that all particle images are
sharp. The magnification of the flow image should be carefully determined, usually by
recording a calibrated scale along with the flow field. The time separation between laser
With a Gateway P5/90 computer and a DT3852 frame grabber, 128 x 128 auto-
correlations take approximately 0.4 seconds per vector, using the 16-bit version of
VISIFLOW. The faster host processor (Pentium 90 MHz) can be as fast as the slower
computer, using an array processor.
With a Gateway P5/133 computer, 128 x 128 auto-correlations take approximately 0.15
seconds per vector, using the 32-bit version of VISIFLOW.
With a P5/90 computer and an Alacron FT200-16-2 200 MFLOPS array processor, 128
x 128 auto-correlations takes approximately 0.05 seconds per vector.
16.1 Introduction
OLE Automation is an industry standard that applications use to expose their OLE
objects to development tools, macro languages, and other applications that support OLE
Automation. For example, a spreadsheet application may expose a worksheet, chart,
cell, or range of cellsall as different types of objects. A word processor might expose
objects such as application, document, paragraph, sentence, bookmark, or selection.
VISIFLOW exposes several objects to a controller. The main object, mapping to a .PIV
datafile has a sub-object, the Application object, which lets you set up several global
parameters, which affect all subsequent datafiles.
For each object, there are two main features that VISIFLOW exposes to a controlling
application; methods and properties. Object methods are functions (procedures) with
calling parameters, that VISIFLOW goes away and performs, possibly returning a value
to the calling program. Object parameters can be assigned to/from variables in the
calling program.
These are subject to change as VISIFLOW develops. Please use the online help to get
the current list.
Main methods
AcquireImage()
AddVector()
AnalyseImage()
Close()
ComputeScalar()
ExportTextAs()
FillGaps()
FinishAnalysis()
GetImageInfo()
GetVector()
LoadSequence()
MaskBoundaries()
OpenBoundary()
PrepareForAnalysis()
Save()
SaveAs()
SetImageInfo()
ShowSequenceNumber()
SmoothData()
UpdateStats()
ValidateData()
AnalysisStartTime
Application
AverageAngle
AverageVelocity
Description
FileName
GapsMinX
GapsMinY
GapsNumX
GapsNumY
GapsVectorSeparation
ImageData
NumberVectors
Application methods
GetCorrelationMasks()
GetFilterValues()
SetCorrelationMasks()
SetFilterValues()
Application properties
AnalysisMethod
ArrayProcessor
BatchAnalysisIncrement
BatchAnalysisNumber
BinariseImage
BinariseThreshold
CorrelationOverlap
CorrelationXRes1
CorrelationXRes2
CorrelationYRes1
CorrelationYRes2
FrameGrabber
GapsFitPower
GapsFitRadius
ImageSource
MicronsPerXPixel
ImageData methods
None at present
ImageData properties
BatchSize
Height
PhysicalHeight
PhysicalWidth
Width
The following examples have been written using MS Visual Basic for Applications, but
they should be readily portable to other control languages.
The following code extract is an example of accessing a previously saved .PIV data-file.
Sub Example1()
Dim VISIFLOW As Object
Set VISIFLOW = GetObject ("C:\VISIFLOW\PIV_DATA\FLOW1.PIV")
VISIFLOW.UpdateStats
iNum = VISIFLOW.NumberVectors
VISIFLOW.Description = "Changed by OLE"
VISIFLOW.SaveAs ("C:\VISIFLOW\PIV_DATA\OLE.PIV")
Set VISIFLOW = Nothing
End
This example loads the FLOW1.PIV datafile, gets the number of vectors into a variable,
changes the datafile description, and then saves it to a different file, OLE.PIV.
Sub Example2()
Dim VISIFLOW As Object
Set VISIFLOW = CreateObject ("VISIFLOW Datafile")
Ver$ = VISIFLOW.Application.Version
VISIFLOW.Application.BatchAnalysisNumber = 5
VISIFLOW.Application.AnalysisMethod = 3
VISIFLOW.Application.CorrelationXRes1 = 64
VISIFLOW.Application.CorrelationRes2 = 64
VISIFLOW.PrepareForAnalysis
lImageData =
VISIFLOW.GetImageInfo("C:\VISIFLOW\PICS\2_000.TIF")
iNumberFiles = VISIFLOW.ImageData.BatchSize
iImageWidth = VISIFLOW.ImageData.Width
iImageHeight = VISIFLOW.ImageData.Height
For iFile = 0 To iNumberFiles - 1
This example creates a new, empty datafile, sets VISIFLOW to use cross-correlation
(64x64 areas), and batch analysis of 5 image files. The example then specifies the base
filename of the batch, and begins a loop over the image files. For each image file, it is
first acquired (loaded from disk), then analysed, then saved. Then a new datafile is
created, and the process repeated for it.