||Structured Light in Matlab
||Olsen, Mikkel Damgaard
||Aanæs, Henrik (Image Analysis and Computer Graphics, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Dahl, Anders Lindbjerg (Image Analysis and Computer Graphics, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
||Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark
||A structured light scanner can be used to reconstruct a 3D-model of an object.
This is possible by use of stereo vision and structured light which can be constructed from the knowledge of a special kind of binary code called Gray code.
Matlab code has been obtained which can be used to reconstruct the 3D-model,
but the code uses a lot of time to do reconstruction. The output of the code is
a point-cloud which is saved in a matrix where the x- y- and z-coordinates are
saved. As the data is saved in Matlab, the point-cloud can only be rendered in
Matlab, which implies a number of limitations. As the scanner is dependent on
flection of light a number of restrictions about the object is inevitable.
In this report the Matlab-code has been optimized in order to perform the 3D-reconstruction faster and the optimization is described in details. A number of
tests has been made in order the see how the scanner reconstructs different kinds
of objects and in order to see how accurate the scanner reconstructs the model of
the object. It has been shown that the scanner can not reconstruct transparent
ective objects, as these surfaces will re
ect the light in an unwanted way
and the structured light will not be determined correctly during the scan. The
same case is apparent for dark objects, as the structured light consist of black
and white stripes and thus the white and black stripes can not be distinguished.
In order to increase the appearance of the model, a number of methods has been
described and implemented in Matlab. These methods use back-projection in
order to obtain the colors of the 3D-points in the point-cloud. The functions
focus on rendering in Matlab and due to this the resulting colored model is slowly
rendered. The technique of texture-mapping has been used to map texture to
the surface of the reconstructed point-cloud and the reconstruction of the surface
has been possible by use of the application MRFSurface. As an ending of the
work with Matlab, a user-guide has been made which can be understood without
knowledge of the theory described in the report.
||Technical University of Denmark (DTU) : Kgs. Lyngby, Denmark
Creation date: 2010-06-25
Update date: 2010-06-25