Non-Local Means with Steer Search Window and Adaptive Parameter

The non-local means(NLM) algorithm is an effective image denoising method, and many variants have been proposed in NLM. However, the search window in most variants is always square which is highly likely to introduce irrelevant sample pixels in denoising. To this end, motivated by steering kernel, an elliptical search window that is defined along the local edge structure is adopted, as well as the strategy of adaptive patch size and smoothing parameter method to improve the ability of noise suppression and detail preservation. Experimental results show the effectiveness and better performance of the proposed method when comparing with other methods.