Analysis of Various Issues in Non-Local Means Image Denoising Algorithm

Abstract: Denoising is still a fundamental, active and unsolved problem in image processing which affects the various high level computer vision tasks like image segmentation, recognition, and tracking etc. The basic goal of denoising is to estimate the original signal from the noisy observations while preserving the important details such as edges and textures. There is always a trade-off between the noise reduction and preserving the important image details. A wide collection of image denoising techniques have been proposed to deal with the denoising problem, but there is still requirement of improvement in the algorithms to enhance the performance of the algorithms. In recent years, the patch based image denoising algorithms like Non-Local Means (NLM) have drawn much more attention to tackle the denoising problem. This paper highlights the various issues of NLM algorithm and presents a review of significant contributions by the different authors to improve the performance of NLM image denoising algorithm.

Keywords: denoising, Non Local Means algorithm, Gaussian noise, peak signal to noise ratio /P>