Super resolution processing of SAR images by Matching Pursuit method based on Genetic Algorithm

This paper adopts Genetic Algorithm in reducing the complexity of the Matching Pursuit method in super resolution of SAR images. Firstly, we introduce the attributed scattering model according to the target feature of SAR images. Secondly, we introduce the sparse representation of image based on Matching Pursuit. This sparse representation evolves into a nonlinear optimization problem, which has a huge computational burden. In this part, we also introduce the structure of the Fourier dictionary, from which we search for the atoms to decompose the images. Thirdly, we introduce Genetic Algorithm to solve the nonlinear optimization problem to get a faster computational speed. Then we apply Genetic Algorithm in searching the best atom from the Fourier dictionary for each step of Matching Pursuit. The reason is given why it is possible to get the faster speed via Genetic Algorithm by analyzing the character of Genetic Algorithm itself and its relationship with the problem theoretically. Finally, we program to realize this idea. The computational results of measured MSTAR data demonstrate that Genetic Algorithm match this problem well. Thus the Matching Pursuit method based on Genetic Algorithm is really effective and fast in SAR image super resolution.