Multiple-input transconductance CMOS amplifier for implementation in continously programmable cellular neural networks

In the paper a concept of reprogrammable Cellular Neural Network realized with multiple-input transconductance amplifiers is presented. A methodology of the circuit design is given, based on the CMOS VLSI technology. This approach is especially attractive for realization of CNNs for real-time image processing since it enables controlling the system templates parameters required during the processing task. The theoretical considerations are illustrated by computer simulation of connected component detector and by measurements of the characteristics of manufactured amplifier