Extracting depth information from stereo linear images using a genetic approach

Stereo vision is one of the most important topics in computer vision. The goal is to compute depth information of a scene seen by two or more video cameras from different viewpoints. The key problem consists of identifying features in stereo images that are generated by the same physical feature in the three-dimensional space. In this paper we present a genetic approach to the stereo correspondence problem where a new solution encoding is proposed. To evaluate a solution, the fitness function is defined from three competing constraints, such that best matches correspond to its minima. Experimental results are presented to demonstrate the effectiveness of the proposed approach for localizing moving objects of a scene seen by a linear stereoscopic sensor.