Circular-Vector Based Non-Local SAR Image Denoising Algorithm

A circular-vector based non-local SAR image denoising algorithm was proposed. Circular vectors were abstracted according to the main direction of the central pixel. Feature vectors were then calculated and the weights of similarity were obtained based on these feature vectors. The execution time and quality of rotation invariance were tested. The results show that the time complexity is significantly lower than the patch-based similarity matching algorithm in NL-Means, and the rotation invariance is also maintained. Experiments have verified that the proposed algorithm shows better similarity matching results comparing to NL-Means and competitive denoising results on PSNR and SSIM against currently state-of-the-art denoising algorithms, such as BM3D, BLS-GSM.