Current mode cellular neural network with digitally adjustable template coefficients

A current mode implementation is presented for a cellular neural network (CNN) with digitally controllable coefficients. The controllability is achieved by applying proper bias voltages to cascode current mirrors. The test circuit is a 1/spl times/5 CNN fabricated in a 1.2 /spl mu/m CMOS process. Measurement results are reported for connected component, shadow and holefiller templates.