Research Of Image Segmentation Based On Pulse Coupled Neural Network

Pulse Coupled Neural Network(PCNN)model is a mathematical model which is based on the phenomenon that the neural cells in the visual cortex of the brain generate synchronous pulse oscillation when they are stimulated by external stimuli,and it is an artificial neural network model which is very close to the real visual neuron.The PCNN model can ignore the small gray difference between the pixels in the same region and make up the gap in the image,so it is widely applied in image segmentation.Aiming at the problems of low efficiency and inaccurate in image segmentation based on the traditional PCNN model,the following work has been done in this paper:Firstly,the basic principle of traditional PCNN is introduced in detail,and the significance of the corresponding mathematical model is briefly expounded.From the experimental results,the image segmentation effect is analyzed,and the deficiencies are pointed out for the model,that is the number of parameters in the model needed to be set is too many,and the efficiency of segmentation is very low by setting parameters by manual multiple tests,Secondly,aiming at the problems that the low efficiency of setting parameters and inaccurate segmentation of traditional PCNN,the mode is simplified,the number of model parameters is reduced,and the setting of the parameters is adaptive to the image grayscale change and its spatial distribution,so the efficiency of image segmentation is improved greatly.The experimental results show that the simplified model can improve the quality of image segmentation,the outline of target is very complete,and the texture is clear.Finally,aiming at the problem that the segmentation of texture in the low contrast image is not accurate with the simplified PCNN model,an improved PCNN model was proposed.In this improved model,the setting of threshold is simplified to improve the efficiency of image segmentation much further,and by increasing the contrast of the image gray,the segmentation accuracy is improved.The experimental results show that the contour of the target is complete,the texture detail is accurate.Compared with the classical Otsu method,Random walk method based on graph theory and Level set method based on energy function,the improved PCNN model method is more effective in image segmentation.