The Research Of The Solution Of L_k Norm Regularization Model For SAR Image Processing

Synthetic Aperture Radar (SAR) is an important approach to get reconnaissance images and remote sensing images. But its images are always degraded by varies of causes, such as noise, speckles or under-sampling.Regularization method, which is one of significant tools to improve the quality of SAR images, builds its model according to the prior information of the original SAR image and the observed SAR data, so we can get SAR image which is quality improved by solving the regularization model built before. Constructing an iterative expression to approximate the exact result of the model is a general idea to work out the model, but its computation is very large, at the same time, the determination of the regularization parameter is a puzzle.According to the sparse prior information of SAR images, we analyze the iteration formula of the solution method of the regularization model based on the l_k norm, and simplify it, then we can achieve the analytical expression of the model's solution. Numerical analysis has validated the feasibleness of the simplification and the correction of the analytical expression mentioned before.We theoretically analyze the choosing method of the regularization parameter which refers to the parameter determination method of general ridge estimation model. Basing on the minimum mean-square error criteria, we find the regularization parameter, then referring to the idea how to get the analytical expression of the regularization model, we can obtain the parameter adaptive regularization model and the analytical expression of its solution.According to the analytical expression of the model solution, we analyze the influence of resolution caused by the model, and then simulated and actual data test results are presented and analyzed.Furthermore, consulting the idea that to find the resolution directly by obtaining its analytical expression, we study the solution of the SAR phase history domain data sparse representation way based on its scattering model, and get the process of its theoretical solution.