Integration of sensor/processor under cellular neural networks paradigm for multimedia applications

Abstract

Compact, high computing power systems become feasible with significant progress in the research and development of advanced computing architecture and array processing. A scalable image sensor array processor with frame memory buffer and cellular neural network (CNN) for nearest neighbor interaction has been developed in a 0.5 mm HP CMOS technology. A CNN with analog programmable weights was constructed with compact mixed-signal circuit components in the current-mode technique. The low voltage, low power operation is supported with the current mode scheme which scales appropriately with the supply voltage. Design of a variable gain neuron circuit can be incorporated into the prototype to realize the optimal solution capability using hardware annealing.