Research On Edge-preserving Filtering Algorithm For Image Salt And Pepper Noise

In the process of acquisition,transmission and storage,images are often polluted by various kinds of noise.When the image is polluted by salt and pepper noise,the damaged pixels take the maximum value and the minimum value,which show as the random distribution of white and black spots on the image.Salt and pepper noise not only seriously reduces the image visual quality,but also causes the failure of subsequent processing such as edge extraction and segmentation.Edge detail is an important part of image content,and also an important basis for image segmentation and feature extraction.It is an important task of salt and pepper noise filtering that how to better restore the edge and other details in the process of denoising.While removing salt and pepper noise,it is difficult to restore the edge details of the image.Especially in the case of high density salt and pepper noise concentration,the restored edge usually has the phenomenon of sawtooth,blur,concave convex,burr and so on.In recent years,most of the research work is focused on the reconstruction of the gray value of noise points,and various types of filters are constructed.Based on the study of the existing typical and improved salt and pepper noise filtering algorithm,this paper proposes some effective methods to better restore the edge details.(1)An adaptive adjustment method of filter window size based on the number of signal points is proposed.A larger window size can make enough signal points participate in the filtering calculation.However,if the window size is too large,the signal points in a long distance will participate in the filtering calculation,thus reducing the accuracy of gray reconstruction.In this paper,on the basis of sufficient experimental tests,the minimum number of signal points in the filter window is determined,which effectively improves the detail retention ability of the filter.(2)A method to calculate the gray weighted mean value of signal points based on edge direction deviation angle and distance is proposed.Firstly,the improved adaptive weighted mean filtering algorithm in the fourth section of this paper is used to remove salt and pepper noise from image,and then the extended Prewitt filter template is used to calculate the image edge direction based on the preliminary denoised image.Finally,based on the edge direction deviation angle of the signal points and the distance to the center of the window,the weights of the signal points in the window are constructed for the mean filtering calculation.The algorithm makes the gray value of the reconstructed noise points more dependent on the adjacent signal points in the direction of the minimum gray change rate,and significantly improves the gray reconstruction accuracy of the noise points.(3)An objective evaluation method of edge similarity is proposed to measure the effect of image edge preservation.The objective evaluation of image quality is to study the perceptual fidelity between two images.By combining the edge direction and amplitude of the image,the edge similarity is evaluated comprehensively.The simulation experiment is carried out by using MATLAB.The performance of the algorithm is evaluated subjectively and objectively.The results show that the salt and pepper noise edge preserving filtering algorithm proposed in this paper can effectively remove the salt and pepper noise and better maintain the edge information of the image,and overcome the edge burr and fuzzy effect brought by the traditional algorithm.