The BJT-based silicon-retina sensory system for direction- and velocity-selective sensing

Abstract

In this paper, a BJT-based silicon-retina sensory system for direction- and velocity-selective sensing is proposed and implemented. This detector can selectively detect the specified direction and speed of moving object image in 2-dimension. The basic detection cell consists of one BJT-based silicon retina cell, one current-mode edge extractor, two delay paths, and four correlators. The signals which are used to be correlated are binary pulsed signals. The BJT-based silicon retina with the current-mode edge extractor can be easily incorporated into various motion measurement systems and associated image processing neural network to reduce the hardware complexity for VLSI implementations and provide robust edge detection. Using the proposed architecture, an experimental 2-D 32x32 motion measurement chip has been fabricated by using 0.6 mu m CMOS technology. The functions of the proposed sensory system are also verified through measurements.