Trinocular stereovision by generalized Hough transform

In this paper we generalize the Hough transform to match the edge segments in trinocular stereovision and determine the parameters of the segments in 3-D (three-dimensional) space. We show that the corresponding segment triplet candidates can be detected by a generalized Hough transform in the parameter plane (theta, phi) which characterizes the 3-D segment orientation. These triplets can then be verified and the position parameters of the 3-D segments can be detected by a Hough transform in the parameter plane (Y, Z). So the matching of geometric primitives in trinocular stereovision images can be found by the cascade of searchings in two 2-D parameters spaces only. Experimental results are satisfactory. Our method shows the following advantages: (1) trinocular stereovision image matching is transformed into searching in 2-D parameter spaces, which much reduces the computational complexity. (2) Matching can be carried out completely in parallel. (3) No a priori similarity between images is needed, so very different views can be used, which improves the precision of 3-D reconstruction. (4) It is very efficient to solve false targets. (5) Our method gives good results even for partially hidden segments. Copyright (C) 1996 Pattern Recognition Society.