Abstract
We propose cortical mechanisms underlying the figure-ground determination that is an essence of surface perception. The underlying mechanism for figure-ground segmentation has not been clarified, although neurons in monkey's V2 show the selectivity to figure. A recent physiological study by H. Zhou, H. S. Friedman, and R. von der Heydt [1] has shown that the activities of neurons in V1, V2, and V4 correspond to figure/ground, a number of which are co-selective to the contrast configuration of stimuli. Previous models have utilized T-junction and ownership-junction for the determination of figure. We propose a netword model for determining figure/ground based on the combination of contrast information.
