On the choice of the parameters for anisotropic diffusion in image processing

Abstract

Anisotropic diffusion filtering is highly dependent on some crucial parameters, such as the conductance function, the gradient threshold parameter and the stopping time of the iterative process. The various alternative options at each stage of the algorithm are examined and evaluated and the best choice is selected. An automatic stopping criterion is proposed, that takes into consideration the quality of the preserved edges as opposed to just the level of smoothing achieved. The proposed scheme is evaluated with the help of real and simulated images, and compared with other state of the art schemes using objective criteria.