Research Of Image Segmentation Based On PCNN And Level Set Method

Image segmentation is an component of image process in image project. It’s the base of following image analysis and image understanding technique, since the problem has been proposed it attracting wide attention and is also research hotspot now. Image segmentation for grey image is concerned in this paper. Two kinds of segmentation approaches are discussed, including segmentation based on PCNN algorithm and level set method.Pulse synchronization phenomena are found in the process of mammalian visual cortex, puts forward the concept of pulse coupled neural network. By imitating the biological neural cells, the model has pulse coupled, similar pulse synchronization properties and can well simulates the fatigue and refractory period of the biological neurons. The paper did further research on the image segmentation based on PCNN, basic model, typical characteristics and action behavior for the PCNN are introduced first. As the model have some shortage using in image segmentation, reducing the parameter need to be set by user and changing descending function to improve the model, using2D histogram and RENYI entropy as criterion to terminate iteration. Compared to maximum entropy and minimum cross entropy improved model, it received a better effect. using2D histogram and RENYI entropy as criterion to terminate iteration. Compared to maximum entropy and minimum cross entropy improved model, it received a better effect.Level set method has widely used in image segmentation, two traditional kinds of level set method GAC model and LIF model is concerned in this paper. As the GAC model is brought boundary leakage easily and LIF model is trapped into local minimum easily, an fusing model is proposed which can automatically adjust the proportion of GAC and LIF component by weight, setting a large weight for GAC model in background region and a large weight for LIF model in edge region.