An Image Denoising Algorithm Combining Global Clustering and Low-Rank Theory

In recent years, image denoising algorithms based on non-local image prior information have been extensively researched. At present, most similar image blocks are gathered by neighborhood block matching, that is, a fixed number of similar image blocks are matched in the field of image blocks by a similarity measurement method, basing on Euclidean distance. Although this method is more efficient, there are some problems: when searching for similar image blocks locally, some special structures, such as image edges or corners that cannot match similar image patterns and existing similar structures in the global scope, are ignored.To solve the above problems, a global clustering method is proposed in this paper to match image blocks, which improves the flexibility of the algorithm. At the same time, to solve the problem of the loss of detailed information such as edge texture when the similar image block matrix is constrained by low-rank, a low-rank sparse image denoising model combined with global clustering is proposed to effectively filter out noise while preserving image details as much as possible.