A non local approach to de-noise SAR images using compressive sensing method

摘要

De noising of Synthetic Aperture Radar (SAR) images has gained a lot of attention lately. Many researchers have proposed different techniques to de noise SAR images among which non local approaches provided better results. Other techniques like graph cut method and compressive sensing method provided significant improvement in parameters used for enhancement of a SAR image. Compressive sensing requires the image to be sparse that is added through the speckle noise in the original image, the image should be compressed before sensing i.e. it is under sampling the image below some threshold value. Further we are applying BM3D filter to de-noise the corrupted image. This paper summarizes the compressive sensing method with a BM3D chosen as Non Local mean. The results are tabulated in comparison to SAR BM3D with different amounts of speckle content.