Modelling space perception in human early vision: a computational approach.

A computational approach in studying vision is important for understanding the biological mechanisms, as well as for integrating the known cortical architecture of primary visual area (V1) and the physiological functions (some of them experimentally proved, some only presumed). We present, in parallel, two models for early vision mechanisms: one for obtaining the disparity map in stereo vision, and the other for edges enhancement in contour integration, both of them seen from such a computational perspective. We implemented the two models and the results of the simulation were consistent with the biological data. They also account for the importance of interactions between parallel channels of visual information processing.