An analog VLSI neural network for real-time image processing in industrial applications

In this paper we present an analog VLSI architecture that implements a neural network for image processing in industrial environment. The analog architecture is highly modular and operates in real time. The circuit implementation is based on simple and effective circuit primitives. Special care has been devoted to the analysis of the linearity and the precision of computation. A test chip, implementing the filtering stage of the architecture, has been designed and realized.<>