Speckle Reduction Algorithms Based On Non-local Means For SAR Images

Synthetic aperture radar(SAR)has been widely used in civil and military fields due to its all-day,all-weather penetrating capability.However,due to its speckle characteristics,the quality of SAR images is contaminated.The speckle of SAR images makes it difficult for the subsequent interpretation of the image.Therefore,the speckle reduction algorithm is one of the important steps in SAR image processing.In the speckle reduction algorithms,the non-local means has attracted great attention because of its better speckle reduction performance,where its block similarity measures and fastness are two important factors,which are essential to improve the performance of the non-local means algorithm.Inspired by the existing literature,this paper studies patches similarity measure and fastness in the speckle reduction algorithm with non-local mean by SAR images.The main work of this paper are as follows:(1)An improved non-local means algorithm based on maximum ratio distance and edge similarity distance is presented.To solve the problem that the existing ratio distance similarity can not measure the block similarity directly,firstly,the sum of the squares of the maximum ratio of the two image blocks is given as the block similarity and the feasibility of the multiplicative model is proved by theoretical inference.However,the similarity measure is more smooth in the heterogeneous region with less difference in the block means,reducing the edge retention characteristics in this region.In order to solve this problem,inspired by the edge methods,an edge similarity distance is given by the Gaussian gamma window to reduce the weight influence in the low contrast region,and to increase the edge retention characteristics of the edge region.Experiments on synthetic images and SAR images demonstrate the advantages of this algorithm over those of existing algorithms.(2)An improved fast non-local means algorithm based on maximum ratio distance integral map is presented.In order to solve the problem of the rapidity of non-local means calculation,the redundancy of the maximum ratio distance is theoretically analyzed.Inspired by the integral map,the maximum ratio distance is introduced into.The computational advantage by the maximum ratio distance integral map is theoretically analyzed.Through experiments on synthetic images and SAR images,the improved algorithm has better performance than those of existing fast and non-fast algorithms.