STEREO VISION BY SELF-ORGANIZATION

We propose a new algorithm for stereoscopic depth perception, where th e depth map is the momentary state of a dynamic process. To each image point we assign a set of possible disparity values. In a dynamic proc ess with competition and cooperation, the correct disparity value is s elected for each image point. Therefore, we solve the correspondence p roblem by a dynamic, self-organizing process, the structure of which s hows analogies to the human visual system. The algorithm can be implem ented in a massive parallel manner and yields good results for either artificial or natural images.