Switched-current techniques for image processing cellular neural networks in MOS VLSI

An architecture and related building blocks are presented for the realization of image processing tasks using current-mode analog-digital circuits. The architecture is based on the cellular neural network paradigm while implementation is made using switched-current circuit techniques. Since just MOS transistors are required as circuit primitives, the proposed circuits are well suited for standard digital CMOS technologies. Also, the sampled-data nature of switched-current techniques allows for easy incorporation of programmability and reconfigurability issues. Empirical results are given for 1.5- mu m N-well double-metal CMOS prototypes.<>