Fast bilateral filtering using the discrete cosine transform and the recursive method

A variety of fast bilateral filtering algorithms have been proposed in recent years. We found that many of them try to decompose the bilateral filter into a series of spatial filters through range kernel decomposition. This paper presents a uniform framework for algorithms in this category. Within this framework, a new accelerating algorithm for bilateral filter is proposed. The proposed algorithm decomposes the range kernel using discrete cosine transform (DCT) and implements the spatial filtering using the recursive method. Experimental results demonstrate that the proposed algorithm outperforms current state-of-the-art algorithms.