Weighted-fitting-based adaptive filtering of images

In this paper, a weighted-fitting-based adaptive method is proposed to smooth images. It is shown both theoretically and experimentally that the proposed method can preserve both edges and corners while it suppresses noise effectively. Compared to the multiresolution adaptive filtering which only adjusts its filter size, the new method can change both the filter size and the filter shape adaptively so that it can smooth complex images effectively. Compared to the recursive adaptive filtering which is computationally expensive and needs to determine the recursive number, the proposed method does not need any input parameter and is computationally efficient