Research On Adaptive Canny Edge Detection Algorithm Based On Multi-Scale Diffusion Equation

The traditional Canny edge detection algorithm is sensitive to noise,the edge pixel location is not accurate enough,and the dual threshold needs artificial setting.To solve the above problems,an adaptive Canny edge detection algorithm based on the multi-scale diffusion equation is proposed.Firstly,a multi-scale adaptive filter is designed and combined with anisotropic diffusion equation to form a multi-scale diffusion equation model.The model uses a multi-scale adaptive filter to perform convolution with the image,reducing the gradient of the larger noise points in the image,and then according to the non-negative decreasing characteristic of the diffusion coefficient equation,it makes the diffusion slows down where the gradient of the image is larger,and the diffusion is accelerated where the gradient is smaller.The smooth denoising of the image is achieved,and the sensitivity of the algorithm to noise is reduced.Then,the idea of linear interpolation is introduced into the process of non-maximal suppression,linear interpolation is used to generate interpolation pixels in the direction of the image gradient,and compared with the gradient value of the central pixel,select the maximum points to improve the positioning accuracy of edge pixels.Finally,the double maximal inter variance method is proposed,and the optimal double threshold value is adaptively selected by using the gray information of the image,which avoids the human intervention in the threshold selection process,and realizes the complete self adaptability of the threshold selection.In order to verify the advantages of this algorithm,we compare and analyze the classical canny and other improved edge detection algorithms in the noise-free and noisy environment respectively,and the results show that the algorithm can achieve quite even better than other algorithms in the noise-free environment.In noisy environments,it can not only effectively remove the high-density pulse signal,but also can keep the edge information of the target object well,and has better noise resistance and stronger edge positioning performance.