AN ALGORITHM FOR STEREOSCOPIC MATCHING WITH QUASI-DOG DISTRIBUTION AS PRIMITIVES
一种以准DOG分布为基元的体视匹配算法

In this paper, a new algorithm for stereoscopic matching is described and some results are presented, The algorithm has following characteristics:(1) quasi-DOG distributions are taken as primitives;(2) constrained by compatibility, uniqueness and ordering; ( 3 ) it is intrinsically a coorperative algorithm.Both theoretical analysis and applied computation have shown that the algorithm is reasonable and effective. It is also supported by the recent psychophysical experiments. Supported by National Science Foundation. Present address: School of Optometry, U.C. Berkely, University of California U.S.A.