Cellular neural network in image filtration tasks

The mathematical model of cellular neural network (CNN) working in discrete time is considered. The possibility of such devices for computation of the convolution of two-dimensional digital signal with a given kernel is shown. The algorithm of construction of a difference equation with constant coefficient, which has the solution of the demanded convolution is given. The hardware implementation of this equation is given the structure of CNN. Model of an exponential smoothing filter of two-dimensional digital signal is considered.