A silicon model of early visual processing

Abstract

An analog model of the first stages of retinal processing has been constructed on a single silicon chip. Each photoreceptor computes the logarithm of the incident light intensity. A resistive network is used to compute a spatially smoothed version of the receptor outputs. An amplified difference between the receptor signals and their smoothed counterparts forms a second-order spatial filter. Measured outputs from an experimental 48 × 48 pixel array show many of the characteristics of the bipolar cells in vertebrate retina.

Additional details