GVF Implementation of Multilayer CNN and Its Application to Image Segmentation

An implementation method of GVF field using multilayer cellular neural networks (CNN) is proposed, which is combined with expanded and thinned model of CNN to realize the image segmentation strategy of active contours. The CNN has parallel processing ability. Therefore it not only solves the problem of computational comlexity of GVF field using traditional serial computation, but also avoids the local minimum in CNN active contours. Guided by GVF information, external image will be evolved until it reaches the desired position. The experiment results show that the proposed method is better than that of Vilarino, et al ,for different input images and it also has real time processing speed.