An Applied Research Of Pulse Coupled Neural Network For Image Processing

Pulse Coupled Neural Network(PCNN)as the third generation of artificial neural network model has a profound biological basis.Different from the traditional artificial neural network models,PCNN is based on biological visual cortex neurons produce synchronized pulses.The pulse shows phenomena of pulse emission.It is more close to the human visual system,and it can be applied in every aspects of image processing.In this paper,we study the problem that the image edge information is not clear and the parameters of PCNN model are difficult to be determined in the field of image segmentation.The main work is as follows:1.Based on the analysis of the basic principle of PCNN and its characteristics,the simplified model of the PCNN is analyzed in detail.According to the salt and pepper noise denoising problems,combining the PCNN model with the IMF algorithm,I propose a method to remove a PCNN based on the combination of IMF and salt and pepper noise.The filter is designed into three parts:(1)position of salt and pepper noise is pointed by the synchronous pulse distribution characteristics of PCNN;(2)using IMF algorithm to calculate the weight of the region;(3)the salt and pepper noise is updated by weight.Comparing with mean filter,median filter,PCNN and median filter,this algorithm in subjective and objective evaluation index show that the algorithm has good ability to eliminate noise and protect the details of image edge feature.2.Aiming at the problem that the PCNN model is difficult to select appropriate network parameters in the field of image segmentation,this paper proposes an adaptive image segmentation method to set the parameters automatically,which uses the features of quantum immune genetic algorithm to optimize global parameters and characteristics of image to realize automatically determining the parameters of the dynamic threshold parameter,the attenuation coefficient and the connection coefficient.In this paper,we use the maximum entropy threshold method,the Otsu method,and the PCNN segmentation method to evaluate the regional consistency and regional contrast.The proposed segmentation method achieves good results and improves different types ability for image processing in comparison with traditional algorithm.And evaluation results are 0.9936 and 0.4480 on the regional consistency and region contrast.The results show that the algorithm has a significant effect on image segmentation.