Three Dimensional Cellular Nonlinear Networks (3D-CNNs) Based Image Processingwith Complex Patterns and Nonlinear Dynamics with oscillations
Keywords:
Nonlinear dynamics; CNNs based chaos; RD-CNNs; PWL functionAbstract
Cellular Neural Networks (CNNs) represent one of the most effective computational
frameworks for modeling complex spatiotemporal pattern formation in image processing systems. The
dynamic behavior of CNN processing units is formally governed by systems of nonlinear differential
equations, which articulate the state evolution of individual cells within the network topology A
specialized variant, Reaction-Diffusion CNNs (RD-CNNs), extends this paradigm by incorporating
reaction-diffusion dynamics inspired by biological systems. This architecture demonstrates remarkable
emergent properties, including the spontaneous generation of spiral wave patterns and autowave
propagation through locally coupled processing units. Such phenomena enable RD-CNNs to simulate
self-organizing patterns observed in natural systems, such as chemical oscillators and neural tissue
