A New Genetic Encoding for Edge Point Stereo Matching

Abstract:- Stereo vision is a well-known technique for obtaining depth information from two or more video images of a scene. The heavily investigated problem in this approach is how to find the corresponding features extracted from the stereo images. In this paper, we present a genetic scheme to the stereo correspondence problem where a new solution encoding is proposed. The problem is formulated in terms of finding pairs of true matches that satisfy local constraints and global ones. Local constraints are used to encode only compatible matches, whereas global constraints are used to define the fitness function such that best correspondences correspond to its minima. To demonstrate its effectiveness, the proposed genetic stereo matching is applied for extracting depth information of a scene seen by a linear stereoscopic sensor