Edge Detection And Analysis Based On Prior Knowledge

With the development of science and technology,image processing has been applied more and more in real life.Especially in recent years,the resurgence of deep learning makes image processing and deep learning closely combine,which solves more and more hot issues.Edge detection plays an important role in image processing.As one of the basic features of the image,the edge contains important information of the image,which distinguishes the target region from the background region.We can extract the objects we are interested in through the edge features.The effect of edge detection will directly affect the subsequent image segmentation,target recognition and other image processing.Therefore,how to effectively extract edge information from images has become a key problem to be studied.In this paper,we propose a new variational model for edge detection based on the Modified Mumford-Shah(MMS)model.The new model is called as Prior Shape Mumford-Shah(PSMS)model.The core part of the model is to establish the prior shape library,also known as the dictionary,and then automatically detect and extract one or more prior shapes from the given dictionary.When the test image is contaminated,missing and mixed with large noise,the method can effectively complete and detect the edge shape of the target to be detected.In addition,the Alternating Direct ion Multiplier Method(ADMM)is used to solve the model,and the convergence of the algorithm is proved theoretically.Numerical experiments not only focus on synthetic images,but also process natural images.A large number of numerical experiments show that this algorithm can obtain ideal edge detection results when dealing with images with missing information,diverse noise types,complex background and multiple targets.Experiments show that the PSMS model has the advantages of good edge detection effect and strong robustness.