Research On Speckle Reduction Algorithm Of Gaofen-3 SAR Image Based On Structured Information Detection And Shear Wave

GF-3 satellite is China's first C-band multi polarization high-resolution SAR satellite.It can deliberately observe land and sea all day and all-weather.And GF-3 satellite has the characteristics of high imaging resolution,large imaging width ratio,high radiation accuracy and multiple imaging modes.No matter military or civil,GF-3 satellite embodies important value.It provides services to the State Oceanic Administration,the Ministry of Civil Affairs,the Ministry of Water Resources,the China Meteorological Administration,etc.And it provides many users with high-quality and high-precision stable observation data.China's high-resolution SAR images no longer rely on imports.However,the imaging mechanism of GF-3 satellite is different from ordinary optical images.It's coherent system makes the image produce unavoidable speckles.The presence of specks greatly reduces the due value of the image itself.The interpretation of SAR image is seriously affected in the later stage.So despeckling becomes an important step in SAR image processing.This paper has done in-depth research on how to reduce speckles more effectively for the GF-3 SAR image.The main tasks completed are:(1)The despeckling algorithm of the GF-3 SAR image based on structured information detection is proposed.First of all,the analysis of the GF-3 SAR image shows that the structural information features are obvious,and the information contained is extremely rich,with obvious points,edges,lines and other characteristics.However,the traditional despeckling algorithm does not consider the structural characteristics of the image,so it will not get effective speckle suppression.According to the shortcomings of the traditional methods,the despeckling algorithm based on structured information detection is proposed and tested.The experimental results show that compared with Lee filter and gamma map filter,the proposed algorithm has good speckle suppression and edge preservation.(2)The despeckling algorithm of GF-3 SAR images of Frost filter based L1/2 norm is proposed.Thesis analyzes and verifies the sparsity of GF-3 SAR image.The traditional despeckling algorithms such as Frost filter,Kuan filter and Gamma MAP filter.And this traditional despeckling algorithms don't take into account the characteristics of GF-3 SAR image,and the traditional Frost filter uses L2 norm as similarity measurement,but the L2 norm does not have good sparsity.We choose the L1/2 norm with better sparsity to improve the traditional Frost filter.According to the experimental results,the proposed algorithm has good speckle suppression and edge preservation.(3)The despeckling algorithm based on improved Frost filter and Shearlet for GF-3 SAR image is proposed.Thesis introduces the basic theoretical knowledge of the Shearlet transform,and analyzes its good sparseness.The special sparseness of the Shearlet is just in line with the speckle suppression of GF-3 SAR image with sparseness characteristics.The tool combines the Frost filter with the L1/2 norm with sparseness characteristics,which makes the algorithm have more superior characteristics.The experimental results show that the proposed algorithm not only has good speckle suppression,but also effectively preserves the edge information structure of the GF-3 SAR image.