A silicon retina system for color constancy

A silicon retina is an intelligent vision sensor that can execute real-time image preprocessing by using a parallel analog circuit that mimics the structure of the neuronal circuits in a vertebrate retina. We have designed and fabricated a frame-based, wide dynamic range silicon retina with an active pixel sensor that approximates the logarithmic illumination-to-voltage transfer characteristics by using a time-dependent stepped reset voltage technique. In the present study, we described the silicon retina and its application to robust color detection against changes of illumination. A present model for color constancy is based on Land´s retinex theory and opponent color theory. In this model, an input image is divided into red, green, and blue channels, and received by the image sensors with a logarithmic response. In each channel, the received image is filtered by a Laplacian-Gaussian spatial filter. And then, the opposite color responses, namely redgreen and yellow-blue, are calculated. The specific color can be detected by specifying the area on the opponent color plane which is composed of two opposite color responses. We consider a hardware implementation of real time color vision system employing three wide dynamic range silicon retinas and FPGA.