A study on Parameter Estimation of Bilateral Filter Using Distribution Distance

Abstract

A bilateral filter has high noise removal properties. However, estimation of suitable parameters using an input image is required to obtain a fine denoised image. A method of parameter estimation that uses the distribution distance has been proposed. This method enables optimal parameter estimation of a bilateral filter using the distribution distance and an assumed noise distribution. The parameters are estimated based on the differences between input and output images. However, input images with several small edges present difficulties. In this paper, we propose a method for estimation of parameters the standard deviation of the prior probability noise distribution. We also present the experimental results of the proposed method.