SAR image de-noising based on non-local similar block matching in NSST domain

The traditional image de-noising in transform domain can get good de-noising effects, but it does not use the redundancy of the image information and the self-similarity of the image. In order to make full use of them and get better de-noising results, we propose a new SAR image de-noising method based on the non-local similar block matching in the non-subsampled shearlet domain. Firstly, we divide the image blocks into similar block groups with different characteristics by using the non-local similar block matching method; then, do the non-subsampled shearlet transform to every group and get the high and low frequency coefficients. Soft threshold is applied to the low frequency coefficients. Because the noise mainly exists in the high frequency component and the image coefficients in the transform domain have some correlation, we can define the adaptive threshold according to the correlation between the coefficients. Finally we can achieve the goal of the SAR image de-noising. The experimental results show that the proposed algorithm can keep more details about the original image and make less artificial texture. The proposed algorithm has stronger ability for image de-noising, and better visual effects.