Image denoising with block-matching and 3D filtering

Abstract

We present a novel approach to still image denoising based on effective filtering in 3D transform domain by combining sliding-window transform processing with block-matching. We process blocks within the image in a sliding manner and utilize the block-matching concept by searching for blocks which are similar to the currently processed one. The matched blocks are stacked together to form a 3D array and due to the similarity between them, the data in the array exhibit high level of correlation. We exploit this correlation by applying a 3D decorrelating unitary transform and effectively attenuate the noise by shrinkage of the transform coefficients. The subsequent inverse 3D transform yields estimates of all matched blocks. After repeating this procedure for all image blocks in sliding manner, the final estimate is computed as weighed average of all overlapping blockestimates. A fast and efficient algorithm implementing the proposed approach is developed. The experimental results show that the proposed method delivers state-of-art denoising performance, both in terms of objective criteria and visual quality.

Fast blockwise SURE shrinkage for image denoising

“…Further progress in this area requires a fast segmentation tool that gives robust disjoint partitions. The major difference between the BSS pixel aggregation (27) and that of BM3D [2], [5] is that our aggregations are made with respect to the SURE optimality instead of heuristics. This raises up an interesting question how to improve BM3D using the proposed BSS idea, and we shall explore this direction in future.…”

“…Consequently, these overlapping blocks lead to an overcomplete problem in determining the final estimated pixels, because each initially denoised pixel x k might be in multiple blocks while each block gives one candidate x k via (25). This overcomplete problem can be approached by reestimating the final denoised pixels x k s from all BSS x k s [2]. To simplify discussion, we pretend b x * l is an image of the same size as the noisy image, but with all zeros for those pixels outside of B l , i.e.…”

“…Assume we have performed NLM denoising with NLM-SURE, obtaining an initial denoised image x = { x l } l∈I with its denoising risk SURE( x) including all intermediate terms in (2) and (4). Notice that SURE( x) is actually estimated as the average of all pixelwise SURE (PSURE) map in (9):…”

“…In recent years, the most popular image denoising algorithms include the NLM method [3], [4] and the BM3D method [2], [5]. Both utilize adaptive patch weighting and aggregations during denoising, while NLM denoising is pixelwise whereas BM3D is patchwise.…”

Robust local similarity filter for the reduction of mixed Gaussian and impulsive noise in color digital images

“…1, in which the restoration of a degraded image distorted by mixed noise is presented using filters intended for Gaussian or impulsive noise suppression. As can be observed, the popular, highly effective techniques like Non-Local Means (NLM) [3,4], Block-Matching and 3D Filtering (BM3D) [9] or Bilateral Filter (BF) [52] are unable to suppress the impulses and the final filtering result is of unacceptable quality (see Fig. 1c-e).…”