Image denoising using non-local fuzzy means

Due to the fact that the dissimilarity between the centered and other patches within an image dynamically changes in each denoising iteration, this paper proposes a new non-local image denoising algorithm called the non-local fuzzy means (NLFM). It considers the weights as fuzzy variables and adaptively update their values by solving an energy minimization problem. A new exponential parameter is introduced to the weights, offering a nonlinear mapping to enhance the NLFM´s denoising performance. Two strategies are proposed to solve the energy minimization problem in different parameter settings. Experiments demonstrate that the NLFM outperforms several existing non-local means algorithms in terms of visual quality and quantitative measures.