Optimal parameters for bilateral filtering and SAS image denoising

Bilateral filtering is a robust denoising method widely used for image preprocessing. Due to the varying properties of the noise, the selection of appropriate parameters for bilateral filtering becomes an important task. In this paper we address this problem by analyzing the performance of bilateral filtering using as a quality measure the VRMSE. First, we propose general rules for choosing suitable bilateral filtering parameters. These rules are then refined to an approach that allows one to determine optimal filter parameters if the standard deviation of the additive noise and a ground truth are known. Finally, we apply this approach to SAS images where the problem of a missing ground truth has been compensated by exploiting the similarity within a set of SAS images.