DocumentCode
2067266
Title
Fast phase-based stereo matching method for 3D shape measurement
Author
Li, Dong ; Zhao, Huijie ; Jiang, Hongzhi
Author_Institution
Dept. of Syst. Design Eng., Univ. of Waterloo, Waterloo, ON, Canada
fYear
2010
fDate
25-27 Oct. 2010
Firstpage
1
Lastpage
5
Abstract
Three dimensional (3D) shape-measurement systems combining fringe projection and stereo vision utilize phase for dense stereo matching. However, projection of orthogonal fringe patterns and computation of curved epipolar lines due to lens distortion has made traditional phase-based stereo matching time-consuming. This paper presents a phase-based stereo matching method that takes advantage of epipolar line rectification and transforms bi-stereo geometry into epipolar standard geometry. Using phase-shifting techniques to compute phase, stereo matching is simplified by searching for the closest phase along horizontal epipolar lines. Compared with the traditional method, this method shortened the stereo matching time by 20 percent. In 3D reconstruction of a plane with high flatness, the 3D shape measurement accuracy was demonstrated to be 0.0178mm. Thus, this method is suitable for applications that require fast speed and high accuracy.
Keywords
image matching; image reconstruction; shape measurement; stereo image processing; 3D shape measurement; bistereo geometry; epipolar line rectification; fringe projection; lens distortion; orthogonal fringe patterns; phase-based stereo matching method; phase-shifting techniques; stereo vision; three dimensional shape measurement systems; Accuracy; Calibration; Cameras; Geometry; Shape measurement; Surface reconstruction; Three dimensional displays; 3D shape measurement; epipolar line rectification; fringe projection; phase-based stereo matching; phase-shifting; stereo vision;
fLanguage
English
Publisher
ieee
Conference_Titel
Optomechatronic Technologies (ISOT), 2010 International Symposium on
Conference_Location
Toronto, ON
Print_ISBN
978-1-4244-7684-8
Type
conf
DOI
10.1109/ISOT.2010.5687348
Filename
5687348
Link To Document