Image processing by cellular neural networks with switching two templates

Abstract

Cellular Neural Networks (CNN) were developed by Chua and Yang in 1988. The main characteristics of CNN are the local connection and the parallel signal processing. CNN consists of cells connected each other and they are arranged in a lattice. CNN is applied to the image processing because the its structure is similar to the image data. The performance of the CNN depends on the parameters which is called the template. When the template has a good influence of the processing, CNN can perform complex processing. In this study, we propose switching two templates CNN. The feature of the proposed method is switching two templates by using the maximum and the minimum output values surrounding the cell. We consider that cells are placed in the input image; edge, background, etc. We apply the proposed method to edge detection and investigate its performance.