Improved BM3D for Real Image Denoising

Block-matching and 3D filtering (BM3D) and most of its improved algorithms assume that the intensity of the noise image is known, while the real image does not know the specific noise intensity. Moreover, many noise estimation methods have the problem that the estimated value is smaller than the true value. In this case, there will be a problem that the denoising effect of BM3D is not up to expectations. Therefore, we propose an improved BM3D algorithm. A local noise variance estimation algorithm combined with block matching is added to the traditional BM3D to obtain the noise variance of the target block and let it participate in subsequent calculations. The experimental results show that the improved algorithm solves the shortcoming that BM3D cannot directly denoise the real image. At the same time, the algorithm solves the problem that the overall noise variance estimate of the image is too small, which leads to the poor denoising effect of BM3D.