Image Denoising Based On Improved 3D Block Matching Algorithm

During the formation,transmission and scanning of images,the image quality is often degraded due to external and internal interference,which affects the visual effect of the image,and has a serious impact on the subsequent processing such as image edge extraction,image segmentation and image recognition.Image denoising is the premise of image processing.The purpose of image denoising is to remove noise as much as possible while retaining useful information such as image edges and texture details.Block-matching and 3D Filtering(BM3D)is one of the best denoising algorithms in the current image processing field,but it will have edge ringing effect at the edge of the image when processing the edge and texture information of the image.In order to solve the problem of BM3D algorithm in processing image edges,an improved algorithm is proposed.Firstly,the anisotropic diffusion filtering is performed on the noise image to obtain the smoothed image and the edge image is preserved;Then the smooth region and the edge region of the image are extracted by the Sobel operator.In the block matching grouping stage of the algorithm,the traditional image smoothing region is used.In the BM3D algorithm,similar blocks are searched in the vertical and horizontal directions;for image edge regions,similar blocks are searched in the vertical and edge directions.This avoids erroneous grouping during the block matching grouping phase,ultimately allowing the edge information of the image to be preserved.At the same time,since the traditional BM3D algorithm adopts a fixed-shaped square search window when searching for a neighborhood,when the window does not adapt to the image features,the denoising performance is degraded.The paper uses blocks of different shapes to accommodate the detailed features of the image,which ensures that the matching block contains enough image detail.The experimental results show that the improved algorithm is superior to the traditional BM3D algorithm in both visual effects and objective numerical indicators.The number of similar blocks obtained is 4 times that of the traditional BM3D algorithm,and the peak signal-to-noise ratio is increased up to 1.29 dB.It is about 0.44dB.At the same time,the denoising performance of the improved algornthm is better than the traditional BM3D algorithm for the image of the contact network with complex edges and the image contaminated by salt and pepper noise.