See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/250355577

CAD Model Generation Using 3D Scanning

Article  in  Advanced Materials Research · October 2007

DOI: 10.4028/www.scientific.net/AMR.23.169

CITATIONS READS

0 423

3 authors, including:

Seyed Yousef Ahmadi Brooghani M. Rakhshkhorshid

University of Birjand Birjand University of Technology
43 PUBLICATIONS   186 CITATIONS    14 PUBLICATIONS   98 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Investigation of Crack Propagation in Carbon Nanotubes View project

Metal Foam View project

All content following this page was uploaded by Khalil Khalili on 09 August 2015.

The user has requested enhancement of the downloaded file.

Advanced Materials Research Vol. 23 (2007) pp. 169-172
online at http://www.scientific.net
© (2007) Trans Tech Publications, Switzerland

CAD Model Generation Using 3D Scanning

K. Khalili1,a, S. Y. Ahmadi-Brooghani1,b, M. Rakhshkhorshid 1,c
1
University of Birjand, Birjand, Iran
a
[email protected], [email protected], [email protected]
b

Keywords: Surface Reconstruction, 3D Scanning, CAD, Reverse Engineering.

Abstract. 3D Scanners are used in industrial applications such as reverse engineering and
inspection. Customization of existing CAD systems is one of rapid ways to supplying a 3D
Scanning software. In this paper, using AutoLisp and Visual Basic programming languages,
AutoCAD has been customized. Also facilities of automatic scanning of physical parts, in the
domain of free form surfaces, have been provided. Furthermore, possibilities such as, control of
scanner automotive system, representation of registered point clouds, generation of polygon and /or
NURBS model from primary or modified point clouds, have been prepared. Triangulation and
image processing techniques along with a new fuzzy logic algorithm have been used to extract the
depth information more accurate. These, accompanying with AutoCAD capabilities have provided
acceptable facilities for 3D scanning.

Introduction
Three dimensional scanning provides the depth information of all scanned points. The coordinate of
surface points of the object relative to a global coordinate system is extracted. This information can be
stored and displayed in the form of point clouds, polygon surfaces, NURBS surfaces, or even a solid
mode. For extracting 3D information, different range finding techniques are used. A variety of
techniques with different capabilities, limitations, and application are used for 3D scanning [1].
Amongst different techniques, mechanical and optical systems are more common. The mechanical
contact scanning uses Coordinate Measuring Machines (CMM’s). Non-contact laser scanning works
based on either the Time Of Flight (TOF) or other techniques such as triangulation. Touch probes,
typically limit the speed of scanning to 1 to 2 points per second [2]. When it is to measure some
representative points, the speed is not important. However, more and more capabilities required of
manufacturing process, call for more detailed mapping of parts to serve not as a measurement of
general trends, but as an actual check of single part integrity. In these cases laser range finders are used
for capturing more detailed and in higher speed. Typically for a laser scanner a speed of 14400 points
per second is expected [3]. On the other hand laser probes have some difficulties with glossy or rough
surfaces and currently laser scanning is not to compete with micron accuracies of some CMMs.
Combined systems such as [3,4] have been developed to overcome some of problems of mechanical
and optical systems.
Triangulation technique is more accurate and robust than TOF. A variety of laser patterns are used to
illuminate the surface of the object (e.g. dot or line pattern) and to extract the depth of object points.
The accuracy in locating the laser light pattern needs to be improved. In addition to the need for
accuracy improvement scanners need improvement in ease of operation. The current paper reports on
the development of a less expensive scanner with improvement both in accuracy and ease of use. A
new fuzzy logic algorithm is developed to more accurately extract the depth information.

Principles of CAD Models

For modeling of curves and surfaces, a variety of representations are used. The approach used in CAD
must be able to exactly represent all user interested curves and surfaces (regular or irregular), also must
have capabilities such as; ease of use and simplicity of algorithm implementation. Hence, in CAD
systems, Polygon, B-spline and NURBS models, are used for construction of curves and surfaces. B-
Spline curves are constructed from piecewise continuous Bezier curves and in the same way, NURBS

All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the
written permission of the publisher: Trans Tech Publications Ltd, Switzerland, www.ttp.net. (ID: 62.60.180.203-08/12/07,08:57:33)
170 Materials and Technologies

curves are constructed from piecewise continuous Rational Bezier curves. Also surfaces are defined by
generalization of curves definitions. In present work BURBS are used to build up the CAD model, see
[5] for details.
System Design and Development. The system is designed to include ordinary functions of
existing scanners as well as improvement in accuracy and ease of use. It was also designed to be
seamlessly working with a CAD geometric modeler. The software of the system includes a
geometric modeler and a programming environment. AutoCAD was considered for geometric
modeling and its capabilities were exploited using the AutoLisp and Visual Basic programming
languages. The software developed also includes the Data Fusion and Processing (DFP). DFP takes
the x, y, and z information from different channels and sort the points to be fed into geometric
modeler. DFP has a refining facility in which it provides user with facilities to edit individual
points. It also provides the user with choices to regionally or globally edit points. The population of
data can be redefined in this stage. There is an option to automatically reduce data points. The
default data population is the initially chosen intervals of point samplings. The DFP calculates the
change of consecutive points and eliminates redundant data where more than a fixed number of
repetitive z coordinate data are present.
Triangulation technique was used to generate depth information. The hardware consists of a
range finding probe, a moving system, a control unit and a place for data storage. A spot laser light
source and a CCD camera were used to extract depth information. The scanning is performed by
movements in x and y directions.
The Software. Simplicity and being user friendly were the main concern in software
development, as an example, by clicking a single command; "RUN" button, scanning will be
performed automatically and the CAD model will be generated with no/little interference. In case,
requested information is not provided in correct format, an appropriate error message will appear
and guide the user to how to provide the information. Switching, aborting, automatic zeroing and
step jumping are included in device control.
3D scanning software must have capabilities such as scanner control, acquisition, representation
and processing of point clouds, fitting special primitives to point clouds, construction of surface
models, texture and image mapping to surfaces and data management. Also, for reverse engineering
purposes, it must be able to export information to CAD standard formats. By customization
AutoCAD, it is possible to achieve most of possibilities of a 3D scanning software. The current
work employed the facilities existing in AutoCAD. By using Auto lisp and Visual Basic
programming languages and adding a new menu, with the title of 3D scanner to menu bar in
AutoCAD environment, possibility of automatic scanning of parts with free form surfaces is
provided. In this menu possibilities required for manual/automatic scanning are supplied. Manual
control of steps can be selected to take control of individual steps before model construction, some
of which are described in the following section.
The first option of 3D scanner menu is “Scanner Control”. By clicking on this option a new
window will be opened, Motion control is done through this menu. In this window possibility of
selecting different resolution for both x and y axis are provided. Smaller resolution is achievable
through semi-step revolution of each motor. Information such as the envelope of scanning, required
resolution and file name to store data is requested. Before scanning starts, the probe must be home
positioned. This is done by selecting the appropriate menu. Step resolution will be used as default.
The second option of 3D scanner menu with the title of "Primary Points Cloud" shows digitized
points. By clicking on this option a proper message appears and asks the user to whether store or
modify points.
Third option is "Edited Point Cloud". In this option, possibilities, needed for data edition and
reduction is provided. Scanning intervals both in x and in y directions and the length and width of
scanning region are requested from user in. Finally, by clicking on “Enter “button, control points
that are selected or generated by this algorithm are displayed on the screen. Fourth option is
“Surface Construction” that constructs a surface passing through point clouds. Constructed surface
is in polygon model. Using “PEDIT” AutoCAD command, polygon model can be converted to
 Advanced Materials Research Vol. 23 171

second or third-degree NURBS model. Constructed surface can be saved in standard formats, and
can be transferred to other CAD/CAM/CAE software. DFP uses a blackboard policy to convey data
to the modeler; AutoCAD. A text file is used as the blackboard. The scanned data points are entered
the text file. The data is then accessed by AutoCAD for CAD model construction. This policy
makes it possible to use (via a third-party software) a different geometric modeler for CAD model
reconstruction.
Moving Mechanism. Motion can be provided either by a servomotor or by a step motor. The
output torque of stepper motors are typically low. As for current applications high torque is not
required, two step motor were used to move the probe.
To increase the achievable resolution of the final moving system, semi-step motion can be used. This
increases the resolution of the whole system considerably without the need for more resolvable motors.
Using step motors eliminate the need for feedback mechanism, hence simplicity of the system. In this
work, two stepper motors were used. A simple circuit was also built to generate half-step movements in
both directions. The mechanical arrangement was an X-Y stage table with a claimed resolution of 0.001
mm.
Control System. Motion control can be done by programmable ICs or by computer and its
interfaces. In this work printer port was used for stepper motors control. A middle circuit was
required to establish the connection between printer port and motors that was designed and made
[5].
Fuzzy Refinement of Spot Light Image. In general, a light spot has a Gussian distribution. It is
assumed that the position of the spot is as same as the position of the centre of area of the spot. The centre
can be extracted by simple routines within image processing software or by simple calculation of the centre
of area of the binarized or grey level image.
The assumption is not correct where at least one of following situation is present: (i) the
reflection is not uniform for reasons such as different color or texture, (ii) the reflection surface is
not flat, and (iii) light path occlusion. Two main techniques have been used for beam thinning [6].
The fist technique works based on the maximum light intensity and thresholding. The second
technique an appropriate filter such as 2D Laplacian to extract the maximum intensity using a grey
level image. These techniques give satisfactory results in ideal situation. If the situation deviate
from the ideal then the distribution of reflection must be know exactly known to extract the centre
point accurately. Fuzzy logic algorithm may be used for thinning the light spot. Two sets of
information including the intensity of each pixel relative to its neighbors and its distance from the
edge can be used by fuzzy logic to thin out the light spot [7].
An appropriate function for the Gussian light distribution may be defined as:

0 xi , j ≤ p
2[( x − p) /(r − p)]2 p ≤ xi , j ≤ q
 i, j
M ( xi , j ) =  (1)
1 − 2[( xi , j − r ) /(r − p)] q ≤ xi , j ≤ r
2

1 r ≤ xi , j

Where p and q are the minimum and maximum light intensity, respectively and r is the average
of p and q. l1 , l2, l3, and l4 are defined as the distance of pixel xi,j from sides. It is assumed that
each pixel has occupied a portion of the central pixel and its degree of occupation is defined based
on the two sets of abovementioned information. The degree of membership of each pixel to central
pixel is determined. The pixel with the highest degree is selected as the central pixel.

Results and discussion

Figure 1 shows the result of a surface model passing through point clouds, read from a text file.
Figure 2 shows a portion of the same shape by user defined limits. Figure 3 shows the same shape
 172 Materials and Technologies

in a different reduced resolution. Figure 4 and 5 show results of converting the surface model to
second and third degree NURBS model.

Fig. 1 Automatic Surface Model Fig. 2 User Defined Model Fig. 3 Reduced Resolution

Fig. 4 Second Degree NURBS Fig. 5 Third Degree NURBS

AutoCAD is used in many engineering scopes and has high facilities of customization. In this
project, AutoCAD has been customized for CAD model generation, using 3D scanning digitized
data. Possibilities needed for points reorganization and surface construction is provided
automatically. Possibilities for representation, edition and manually removing/reducing of points,
construction surfaces passing through manually/automatically modified points, user defined limits
and resolutions are provided. Also possibilities for scanner control and data acquisition are
supplied. Ease of use, low cost, and improvement in accuracy are the advantages of the developed
system. Despite of facilities provided, this system has some restrictions. It is not possible to
generate a 90 degree draft angle, not in configuration nor in algorithms; negative draft angle cannot
be created. For overcoming these problems, configuration of fixed probe and object rotation about a
fixed axis is proposed.

Conclusions
The results of this proposed configuration are closed curves that by lofting can be used for modeling
of more sophisticated objects. The accuracy of spot light generation is low but it simply can be
changed for a better light source with sharp spot with a smaller size. Models that are constructed
here can be converted to IGES and therefore can be used in CAM systems for simulation and NC
code generating. Other AutoCAD facilities such as image and surface mapping to surfaces and
manually modification of surfaces, convert the offered system to an acceptable 3D scanning system.

References
1. K. G. Harding, “Application Issue When Using Optical 3D Systems In Place Of CMMs”, GE
Research & Development Centre, (2001).
2. Information on: http://www.laserdesign.com
3. V. H. Chan, M. Samaan, Proc. SPIE, The Int. Society for Optical Eng. Vol. 5302, (2004), p. 33.
4. M. Callieri, et al, Proc. 2nd Int. Symp. On 3D Data Proc., Visualization, and Transmission, (2004).
5. M. Rakhshkhorshid, MSc. Dissertation, University of Birjand, Iran, (2005).
6. J. M. Brigini, T. Redarce, and A. Tutard, RAPA, Vol. 6, (1993), p. 43.
7. S. K. Pal, and A. Gosh, Fuzzy Geometry in Image Analysis, (Elsevier Sci. Publishers B.V., 1992).

View publication stats