Abstract：A useful binocular 3-D vision system is introduced in this paper, the pose of an object with 6 degreesof freedom is discussed. Frei and Chen method is used to extract edges with the help of non-equal quantization and preemphasis techniqucs. Modified Moravec's interest operator method is applied to en-sure automatic extraction of feature points. Edge-based sequential registration and hicrarchical matching significantly speed up the matching process. 4 rules summarized from rigid body movement is used in theperiod of locating object pose from 3-D data to solve combinational explosion and ambiguity problemsduring the matching process of workpiece feature points and model points.
1 Jin S, Li Y. Acquiring range data in binocular 3-D computer vision using edge-based hierarchical matching. In Proc IAPR Workshop on CV, Tokyo, Japan, 1988 2 Barnes D I, Silvcrman H E. A class of algorithms Cor fast digital image registration. IEEE Trans Computers, 1972;C-21(2):179-186 3 Wong R Y. Scqucntial scene matching using edge features. IEEE Trans Aerospace and Elcctronis Systems, 1978;AES-14(1):128-140 4 Wong R Y, Hall E G. Sequential hierachical scent matching. IEEE Trans Computers, 1978;C-27(4):359-366 5 Moravec H P. Visual mapping by a robot rover. Proc 8th IJCAI, 1979:598-600 6 Moravc}c H P. Towards automatic visual obstacle avoidance. Proc 6th IJCAI, 1977:584 7 Chen N Y, Birk J R, Kelley P B. Estimating workpiece pose using the feature points method. IEEE Trans Automatic Control, 1980;AC-25(6):1027-1041 8 Faugeras O D, Herbert M. The representation, recognition and locating of 3-D objects. Int Joint Conf Robotics Restarch, 1986;5(3):27-52 9 Ayache N, Faugcras O, Favcrjon B. A geomcric matchct for recognization 3-D rigid object. SPIE Vol. 521 IRCV 1984:153-159 10 Trivcdi H P. A computaional theory of stereo vision. In Proc CVPR, 1985:277-282 11 Frci W, Chen C. Fast boundary detection:a generalization and a new algorithm IEEE Trans Computers, 1977 C-26(10):988-998