A hierarchical machine vision system based on a model of the primate visual system

The problem of advancing machine visual pattern recognition capabilities is approached by examining the visual system of the primate. A model of biological vision is suggested, and an analogous machine vision simulation is developed. The modeling is limited to luminance information (color, motion, and depth are not considered), and biological systems are considered at the network level (biochemical and biophysical details are not simulated). The system architecture consists of a set of invariance transforms (luminance, spatial, and scale) followed by storage using an adaptive resonance theory network.<
>