Adaptive Search Based Non-Local Means Image De-Noising

Non-Local Means (NLM) is currently a leading approach in the image de-noising field. It uses the weighted average of all pixels in an image to recover the noise-free pixel values. However, if the pixel values used for average do not lie in the same gray levels as the one to be restored, they will make negative contributions to the de-noising performance. In this paper, we present an adaptive search based image de-noising method that combines the Non-Local Means algorithm with a locally adaptive anisotropic estimation technique, which can select the "right" pixels for NLM averaging process. Experiment results show that the proposed algorithm has superior performance than the original one in terms of PSNR^ SSIM, and also in visual quality.