Visual Motion Analysis by a Neural Network

Abstract – In the visual systems of mammals, visual scenes are analyzed in parallel by separate channels. Information concerning visual motion is mainly analyzed through the occipitoparietal pathway. In area MST of the pathway, there are cells that respond selectively to specific motions of a large area of the visual field, such as rotation, expansion and contraction. Many of these cells are reported to respond steadily even when the location of the center of optic flow shifts on the retina. This paper proposes a neural network model that explains the position-invariant responses of the MST cells. The network has a hierarchical multilayered architecture and extracts optic flow based on vector field hypothesis.