Morphological Edge Detection Algorithm Of Colon Pathological Sections Based On Shearlet

The edge detection of colon pathology section is the key technology to construct the computer-aided diagnosis system of colon cancer,and the traditional detection method will be unclear and jagged when extracting the edge,which will affect the diagnosis of the disease.Therefore,it makes sense to look for edge detection algorithms that balance edge clarity with noise-cancelling capabilities.Considering the advantages of Shearlet transformation,mathematical morphology and BP neural network in image processing,this paper mainly studies the edge detection algorithm of improved morphology and BP neural network fusion based on Shearlet de-noise.Since Shearlet often choose Meyer wavelet as a basis function,and the characteristics of Meyer wavelet depend on the selection of the Sigmoid function in the Meyer scale function.Therefore,this paper gives a kind of fully smooth Sigmoid function construction method,and gives the corresponding Meyer scale function and Meyer wave function,on the basis of which,using Sigmoid function to construct a broader Meyer scale function and Meyer wave function general expression.Taking the fully smooth Sigmoid function as an example,the corresponding Meyer wave and Shearlet were combined to de-noise the colon pathological section,and good results were obtained.In morphological edge detection,taking into account the different effects of structural elements of different scales and directions to extract image edges.This paper,based on the idea of multi-scale multi-structure,proposes an adaptive morphological edge detection algorithm,which applies the algorithm to edge detection of the image after Shearlet de-noise,and the simulation results show that the detected colon pathological image edge noise is small,the lines are continuous and clear.In BP neural network,the selection of excitation function directly affects the learning ability of BP neural network.The commonly used excitation function is fixed because its expression is fixed,so its position,steepness,and mapping range in the network are also fixed.In order to improve the learning ability of BP neural network,better edge detection results are obtained.In this paper,a class of fully smooth Sigmoid functions is constructed as the excitation function of BP neural network for edge detection.The simulation results show that the method can get clear edge details while effectively filtering out noise,but there are still some shortcomings,such as some edges are not smooth enough.Since each edge detection method has limitations and shortcomings,it is a common method to use the edge detection algorithm with complementary advantages.Therefore,this paper integrates the mathematical morphological algorithm of BP neural network edge detection,obtains the image edge with low noise and rich detail,is beneficial to the subsequent application of computer-aided diagnostic system for diagnosis,and provides a new way for medical image processing.