Templates and algorithms for two-layer cellular neural networks
Zonghuang Yang, Yoshifumi Nishio, Akio Ushida · 2003
Presents two-layer cellular neural networks for some image processing applications, in which two templates are introduced to couple between the two layers. Several simulations such as linear non-separable task, center point detection and skeletonizing, are executed with the two-layer CNN and their templates are given. All of them display that the two-layer CNNs behave more efficiently for image processing compared with single-layer CNNs. In addition, the stability of the two-layer CNN with symmetric templates and/or special coupling templates is also discussed based on the Lyapunov function technique. Its equilibrium points are found from the trajectories in a phase plane. These results agree with those from simulations.