Despeckling Algorithm Of High Resolution SAR Images Based On Structure Information Detection And Partial Differential Equations

Synthetic Aperture Radar(SAR) has been used in the ecology, hydrology, geography, ocean monitoring, terrain mapping and so on widely with virtues of all-day, all-weather, multiple perspectives and penetration. However, Owing to the coherence from radar waves, SAR images is inevitable to produce some speckle when imaging processing, especially high resolution SAR images, those are more complex than common SAR images. The speckle could lower the quality of SAR image, make it difficult for the image interpretation and subsequent processing. Therefore, it has important research meaning to reduce speckle effectively in high resolution SAR images.Proceeding from spatial filters, partial differential equations(PDE) method and non-local means(NLM) filter, further research had been discussed about how to suppress speckle effectively in high resolution SAR images. The accomplished work as following.(1) Classic despeckling algorithms in SAR images had been proposed, including spatial filters such as Lee filter, Kuan filter, Frost filter, enhanced algorithm, Gamma MAP filter and PDE method such as total variation(TV) model and PM model. And relative merits of these classic despeckling algorithms had been analyzed.(2) A despeckling algorithm based on the local structure information detection and the Kirsch directional operator had been proposed. Owing to the different areas features in SAR image had not been taken into account by the classic Kirsch directional operator, speckle in SAR images could not suppress effective. Therefore, the despeckling algorithm based on the local structure information detection and the Kirsch directional operator had been put forward in this paper to solve the problem of adaptive selection of sliding window by combining the point target detection, the technology of adaptive selection of sliding window and the local structure information detection. This despeckling algorithm could distinguish the point target, edge areas and uniform areas in the image effectively, and could adopt different despeckling algorithm methods according to different areas. Through the despeckling experiments, it could validate the proposed algorithms could suppress speckle in homogeneous area fully, at the same time it could keep the edge areas and point targets effectively, achieving a better balance between despeckling and preserving the edge information.(3) A speckle suppression algorithm is proposed based on the NLM filter and the AA model. Owing to the AA model had more effective performance in suppressing speckle in the homogeneous regions, but poor performance in keeping the edge information. On the contrary, the NLM filter had more effective performance in keeping the edge information, but poor performance in suppressing speckle in the homogeneous regions. Therefore, This paper takes the nonlocal dirichlet function as a linear regularization item, which constructs the weight by measuring the similarity of images. Then, a new despeckling model is introduced by combining the regularization item and the data item of AA model, and an iterative algorithm is proposed to solve the new despeckling model. The experiments show that, compared with the AA model and the NLM filter, the proposed model has more effective performance in suppressing speckle, and it also has better performance in keeping the edge information.