An Improved Image Denosing Method Based on BM3D Algorithm
In the field of image processing,the Block-matching and 3D filtering(BM3D)combines relevant characteristics of spatial domain and frequency domain,and it is one of the best image denoising algorithms at present.Because the algorithm has time-consuming computation,inaccurate extraction of similar blocks and inability to extract similar features with different distribu-tions in segmentation blocks,the paper proposes a method of dimensionality reduction of image information based on Radon trans-form,and judges similar block information by solving variance value of dimensionality reduction function and coordinating location of maximum variance.And then the pixel mean ratio of each similar block relative to the reference block is calculated and scaled for denoising.The experimental results show that compared with the original algorithm,the running time of the denoised image is short-er and the image quality is improved.
国家科技期刊平台
NSTL
万方数据
