Color Images Based On Cellular Neural Network Edge Extraction Research

Cellular neural network is a mathematical model which simulate biological function. Like a neural network, it is a large-scale nonlinear analog circuit which processes signals in real time. Like cellular automata, it is made of a massive aggregate of regularly spaced circuit clones, called cells, which communicate with each other directly only through its nearest neighbors. The characteristics of the local interconnect is particularly suited to use VLSI technology.There are broad prospects for development in the field of image and video signal processing. And it is one of the hot research field of artificial neural network. Edge is one of the important characteristics of image, and it also is the element of several research areas, like computer vision and pattern identification. This paper mainly studied the CNN’s image edge extraction in the field of image processing.Firstly, it not only introduced the research and present application situation of Cellular neural networks but also introduced the process of edge based on CNN. At the same time, it studied the theory and model structure of CNN, and discussed the CNN dynamic range and stability. It analyzed the scope of template parameters in binary images and grayscale images,and the results images are given.Secondly, it summarized the advantages and disadvantages between traditional edge detection and CNN edge detection. Then put forward an new CNN edge extraction method based on nonlinear alternation, and the test results show that the method can effectively extract the image edge.Finally, a new method is proposed to extract the color image. The method is named as vector angle of visual perception based on CNN edge. It considered the human visual system which sensitive to the R, G, B and the angle of color difference factors. This method took the factors as color space distance to determine the template parameter, which can realizate the color image edge extraction. And the experimental results show that, this algorithm are more consistent with human visual perception effect than the original algorithm.