Two Theorems on the Robust Designs of a Kind of Uncoupled CNNs with Applications

The cellular neural/nonlinear network (CNN) has become a new tool for image and signal processing, robotic and biological visions, and higher brain functions. The robust designs for CNN templates are one of the important issues for the practical applications of the CNNs. This paper sets up two new theorems for robust designs of a kind of uncoupled CNNs. The two theorems provide parameter inequalities to determine parameter intervals for implementing prescribed image processing functions, respectively. Four examples for detecting edges and corners in images are presented in order to illustrate the effectiveness of the methodology.