Local Filtering Framework Using Sorting Clustering

In view of the problem of blurred edges in the traditional local filtering framework,a local filtering framework using sorted clustering is proposed.First,the local window is divided into 4 sub-windows.And the point to be processed is located at the intersection of the 4 sub-windows to achieve the purpose of maintaining the edge strategically.Then,a sorted clustering algorithm is proposed by sorting to improve intra-class similar-ity and inter-class differences and using the similarity between pixel values in sub-windows.After clustering,only the pixels in the sub-window that are similar to the points to be processed are used for filtering.Finally,the final filtering result is the one with the smallest difference between the 4 filtering results and the point to be processed.This method further improves the algorithm’s ability to maintain edges.The experimental results based on the SSID data set show that the filtering algorithms based on the filtering framework can achieve higher PSNR and SSIM values.And the proposed filtering framework can effectively improve the ability of traditional local filtering algorithms to maintain edges and filtering,and has certain robustness.