Color categorization and color constancy in a neural network model of V4

摘要: We develop a neural network model that instantiates color constancy and categorization in single unified framework. Previous models achieve similar effects but ignore important biological constraints. Color this is achieved by new application of the double opponent cells found "blobs" visual cortex. emerges naturally, as consequence processing chromatic stimuli vectors four-dimensional space. A computer simulation subjected to classic psychophysical tests first uncovered these phenomena, its response matches results very closely.