Medical Image Denoising Based Block Maching

With the improvement of science and technology,related tecnology of medical image processing has been rapid developing,it has become vitally important part in modern medicine science.Its main role is to obtain the more clear and detailed understanding about disease through the analysis of medical image information,which is conducive to the further medical procedures for patients,On the other hand,Medical image can promote the progress of research and development of cell biology.However in the process of collection and transmission of images,images usually inevitably be contaminated by noise,which not only reduced the visual effect of image,but affected subsequent image processing.As the pretreatment process,image denoising has important value in aspect of theory and application.Block matching 3-D(BM3D)algorithm represent the state-of-the-art denoising method,but the original bm3 d algorithm still have some defects,such as complicated parameter settings,poor adaptability,denoising performance relatively depends on the chosen of parameters,in particular need to know the noise variance as input parameter,in addition,the high time complexity also affect the real-time application of the algorithm.In order to solve the above problems,this paper proposed following improvements.Fistly,we get the estimate of noise variance by iteration based on principal component analysis method,use the value as the input parameter of bm3 d algorithm.Secondly,we proposed a fast matching strategy to reduce the time complexity.Finally,we gave a parallel realization of bm3 d basedon Multi Point Interface(MPI)through constructing distributed computing cluster environment.Experimental results show that the proposed algorithm achieved a good result in terms of performance and efficiency,greatly enhanced the practicality of the algorithm.