Abstract：After pointing out some essential difficulties in binocular stereo vision systems, we proposedthen a trinocular stereo vision method matching the feature primitives in 3-D space by use of pyramidalgraph.This method is based on the 3-D perspective projection geometry and needs no prior constraints or im-age similarities,so it can solve better the contradiction between the accuracy and matching difflculty. And theuse of pyramidal graph can speed up the matching process also. The experimental result for artirlcial imagesshows that the ratio Of false targets is almost zero,and the matching speed and accuracy is higher.
[1] Shen J, Castan S. Zhao J. A new passive measurement method by trinocular stereo vision. Industrial Metrology, 1990,1(3):231-259.
[2] Shen J, Castan S, Zhao J. A new trinocular stereo vision method,Proc of 6th Scandinavian Conference on Image Analysis (SCIA'89),Oulu, Finland,June. 1989.
[3] Shen J, Castan S, Zhao J. Stereo vision by pyramidal BLI graph matching. Proc of International Symposium on Optical and Electro-Optical Applied Science and Engineering,Real Time Image Processing Concepts and Technology (SPIE'87),Cannes, 1987: 108-113.
[4] Tsai T. An efficient and accurate camera calibration technique for 3d machine vision,Proc CVPR'86, Miami,1986.
[5] Shen J, Castan S. An optimal linear operator for edge detection. Proc CVPR'86. Miami, 1986.
