Combined Non-Local averaging and intersection of confidence intervals for image de-noising

The Non-Local means (NL-means) algorithm was recently introduced and shown state of the art performance in image de-noising. However the computational complexity of the NL- means is very high which can prevent its implementation into products. In this paper we propose a method which retains the excellent de-noising characteristics of the NL-means algorithm but with much less computational complexity. Comparison between our method, the NL-means algorithm and other fast NL-means based approaches is illustrated through several experiments.