TY - THES

T1 - Development of a Photometric Stereo Measurement System

AU - Jaindl, Georg Erich

N1 - embargoed until null

PY - 2009

Y1 - 2009

N2 - Photometric Stereo is an imaging and image processing technique for contact-free surface measurement. It uses the characteristic surface reflectance of materials under different lighting conditions to separate surface albedo from surface geometry. The surface of a stationary test specimen is successively illuminated by light sources with known positions relative to the test specimen and camera. The varying light intensity of each surface element gives information about its orientation. This information about the surface gradient is used to reconstruct a discrete 3D-representation of the surface, i.e. surface reconstruction from gradient fields. This work describes a fully automated experiment setup for evaluating Photometric Stereo for both ideal and non-ideal Lambertian surfaces. Error sources in the experiment setup were identified and a new calibration method was developed. This method compensates for the inhomogeneous illumination generated by the light sources. For this purpose, the light distribution of each light source was measured and a calibration matrix was derived. The calibration matrix was used to normalize the source data for Photometric Stereo. The error caused by the inhomogeneous light distribution increases with the size of the surveyed specimen. The developed calibration method allows the application of Photometric Stereo on large specimens. It is shown that a square metallic surface with an edge length of 17.5 cm was reconstructed successfully.

AB - Photometric Stereo is an imaging and image processing technique for contact-free surface measurement. It uses the characteristic surface reflectance of materials under different lighting conditions to separate surface albedo from surface geometry. The surface of a stationary test specimen is successively illuminated by light sources with known positions relative to the test specimen and camera. The varying light intensity of each surface element gives information about its orientation. This information about the surface gradient is used to reconstruct a discrete 3D-representation of the surface, i.e. surface reconstruction from gradient fields. This work describes a fully automated experiment setup for evaluating Photometric Stereo for both ideal and non-ideal Lambertian surfaces. Error sources in the experiment setup were identified and a new calibration method was developed. This method compensates for the inhomogeneous illumination generated by the light sources. For this purpose, the light distribution of each light source was measured and a calibration matrix was derived. The calibration matrix was used to normalize the source data for Photometric Stereo. The error caused by the inhomogeneous light distribution increases with the size of the surveyed specimen. The developed calibration method allows the application of Photometric Stereo on large specimens. It is shown that a square metallic surface with an edge length of 17.5 cm was reconstructed successfully.

KW - Photometric Stereo Calibration of Photometric Stereo Photometric Stereo applied to non-Lambertian surfaces Shape recovery

KW - Photometric Stereo Kalibrierung von Photometric Stereo Photometric Stereo angewandt auf nicht-Lambertsche Oberflächen Oberflächenrekonstruktion

M3 - Diploma Thesis

ER -