The Speckle Suppression Algorithms Based On Logarithmic Product Of Ratio And Nearly Half Number Of Integral-image For PolSAR Imagery

Polarimetric synthetic aperture radar(PolSAR)has been widely used in the fields of land and resources survey,environmental monitoring,military reconnaissance,etc,because of its capability to capture images of the earth in both day and night,and for almost all weather conditions.However,the presence of speckle complicates the image interpretation,it is of importance to suppress speckle.Nowdays,there are many speckle reduction algorithms for PolSAR images,in which the nonlocal means algorithm has been widely concerned because of its good speckle reduction performance.However,there are still some problems in these nonlocal means algorithms,such as insufficient point target preserving,reduction of speckle suppression in the homogeneous region near edge,and redundancy in integral-image calculation.In order to solve these problems,this thesis presents a nonlocal means algorithm based on the logarithmic product of ratio and the numbers of integral images nearly half reduction,which improves the speckle suppression of PolSAR images from theoretical analysis and performance evaluation.The main work of this thesis is as follows:1)A nonlocal means(NLM)speckle reduction algorithm for PolSAR images based on log product of ratio is proposed.To solve the problem of infeasibility on pixel similarity in certain existing algorithm,a distance measure based on logarithmic product of ratio is proposed.According to mathematical analysis,it could be demonstrated that this distance is suitable for multiplicative model under the existing assumption.However,the problems in point target preservation and homogeneous region smooth could be seen according to theoretical analysis.In order to solve these two problems,the distance measure of logarithmic product of block center ratio is introduced.Moreover,the theoretical analysis is exploited to verify the feasibility on the integration of above two distances.Furthermore,these two distances are used to form the nonlocal means algorithm for PolSAR imagery.Three kinds of PolSAR images are used in the experiments,and the results show that the proposed algorithm is better than the existing algorithms,at least 0.056 in the point target preservation and 9.1%in homogeneous region smooth near edge.The experimental results show that the proposed algorithm could effectively solve the problems of point target preservation and homogeneous region smooth near edge.Meanwhile,comparisons of the proposed algorithm and existing algorithms based on other performance indices have also been completed.2)An improved nonlocal means algorithm with the nearly half number of integral-image is proposed.First of all,the computational redundancy of the NLM algorithm is theoretically analyzed.Then,the reason that integral-image could be used to reduce the redundancy of the NLM algorithm is analyzed.However,certain redundance still exists.In order to solve this problem,inspired by the existing literature,the symmetry idea is introduced into the integral-image combined with logarithmic product of ratio to form the nonlocal means algorithm.Theoretical analysis indicates that the numbers of integral-image of the modified algorithm could be nearly half,and the time complexity is reduced in comparison with the integral-image based algorithm.Five kinds of PolSAR images are used in the experiments.The results show that the running time of the modified algorithm has the superiority at most approximate 40%to those of the comparing nonlocal means algorithms,demonstrating that this algorithm could effectively reduce the computational burden of the nonlocal means algorithm and improve its computational speed.Meanwhile,the combination of integral-image and symmetry could maintain the performance of nonlocal means algorithm.At the same time,comparisons of the modified algorithm and existing algorithms based on other performance indices have also been completed.