Research Of Image Segmentation Based On Pulse Coupled Neural Network Model

Pulse Coupled Neural Network(PCNN)is a new generation of artificial neural network simulated by mammalian visual nervous system,which is very beneficial for image segmentation and has attracted much attention because of synchronous pulse emission characteristics,spatio-temporal comprehensive characteristics,and automatic wave characteristics,etc.PCNN has become a research hotspot in the field of image segmentation.The main research contents of the thesis are as follows:Firstly,the traditional image segmentation algorithm and the basic theory of pulse coupled neural network are analyzed,the principle and basic characteristics of the pulse coupled neural network model are deeply studied.Secondly,a new idea of combining improved immune genetic algorithm with PCNN model is proposed,and the improved method of immune genetic algorithm is designed and image segmentation experiment.The results show that the immune genetic algorithm improved by the OTSU algorithm is more effective in segmentation of multi-peak,single-peak and gray-value distribution images than the regional consistency and entropy function.The immune genetic algorithm verifies the practicability of the proposed algorithm.Finally,the PCNN image segmentation methods based on OTSU,minimum cross entropy and genetic algorithm are analyzed respectively.These three algorithms improve the loop termination condition and adaptive setting of model parameters of PCNN image segmentation algorithm,but the complexity of the PCNN image segmentation algorithm itself still exists.Aiming at this problem,this thesis proposes to combine the improved immune genetic algorithm with the PCNN model to obtain the IGA-PCNN image segmentation method:firstly,the improved immune genetic algorithm is used to adaptively obtain the optimal threshold,and then the optimal threshold is replaced by the dynamic threshold in the PCNN model,finally,the image segmentation is completed by using the pulse coupling characteristics of the PCNN model.The IGA-PCNN algorithm not only reduces the setting of dynamic threshold related parameters in the PCNN model,but also reduces the complexity of the segmentation process.The experimental results show that the IGA-PCNN image segmentation method has a good segmentation effect on visible light images with multiple peaks,single peaks and uniform gray value distribution.