Design And Implementation Of Color Image Contour Extraction Algorithm Based On PCNN

The contour information is a kind of feature information which is very researchable in the basic features of the image.Extracting the accurate contour of the target object in the image provides great convenience for the subsequent processing of the image.At the same time,as an important basis of feature extraction and target recognition,it has received great attention in the field of digital image processing and computer vision,and has important application value and research significance.Pulse Coupled Neural Network(PCNN)is a kind of bionic pulse neural network.Its biological background and the characteristics of synchronous impulse emission of neurons make it have unique advantages in the application of contour extraction.However,in the process of PCNN iteration,the neurons stimulated by other neurons can only emit pulses in the next iteration,so there are breakpoints in the extracted contour.At the same time,with the progress of science and technology and the popularity of color images,the limitation of PCNN that cannot directly process color images also limits its play.In view of the shortcomings and difficulties of the existing methods for color image contour extraction,this paper proposes an RG-PCNN color image contour extraction algorithm and a kind of three-channel fusion strategy based on the in-depth study of PCNN model.The two algorithms of color contour extraction of RG-PCNN under the guidance of super pixel information are aimed at further exploiting the advantages of PCNN in contour extraction and solving the limitations of its existence.The main work of the whole paper includes:1)An RG-PCNN color image contour extraction under a three-channel fusion strategy is proposed.The main idea is to use RG-PCNN to extract the contours of three grayscale images separately after RGB three-channel separation of color images,and then perform weighted fusion,which solves the problem that traditional PCNN cannot directly process color images.The RG-PCNN combined with the regional growth algorithm and PCNN can solve the breakpoint problem of the contour to a certain extent.2)A color contour extraction of RG-PCNN under the guidance of super-pixel information is proposed.The algorithm introduces the SLIC super-pixel to guide the RG-PCNN in the contour extraction algorithm proposed above.The super-pixel can maintain the characteristics of the complete boundary well and further solve the situation that the contour has a breakpoint.Different experimental results are obtained by adjusting two parameters:the number of super-pixels n and the maximum value?₈(6)of the joint term coefficients in RG-PCNN,and the optimal parameters were set according to the experimental results.3)A large number of experiments were carried out in the BSDS500 dataset,and the results were quantitatively and qualitatively compared and analyzed.The experimental results obtained confirm the validity and practicability of the algorithm.4)Based on Matlab graphical user interface,a color image contour extraction system based on PCNN is designed and implemented.The system has the characteristic of simple and clear interface and convenient operation.It constructs a practical experimental platform for contour extraction.