Research On Speckle Filtering Algorithms Of SAR Image

Speckle degrades the quality of Synthetic Aperture Radar (SAR) images. The performance of speckle will have an effect on interpretation or further analysis and processing for SAR images. Therefore speckle filtering technique is one of the important problems in the application of SAR. It can be classified into three categories: multi-look processing, statistical filtering and filtering in transform domain. This paper focuses on the research of speckle filtering algorithms of SAR image and obtains some useful conclusions. The author mainly finished several following work. Firstly, speckle's basic principles are introduced in detail and some correlative ideas are analyzed systematically. Secondly, the previous work in SAR speckle filtering is reviewed, especially several well-known statistical filtering algorithms widely used are surveyed. Thirdly, an improved algorithm is proposed for the traditional TDRGMAP algorithm. The new filtering technique is highly effective in the preservation of edge and texture of SAR images. At last, a set of criteria are established to evaluation various filtering performances ranging from speckle suppression to edge preservation. The paper analyzes the processing results of real airborne SAR images for every chapter, and shows that the proposed method attains better improved effect. The algorithms' feasibility and validity are validated.