A Concise Review of Fast Bilateral Filtering

During last couple of decades, bilateral filtering has been an important research topic in the computer vision and image processing community. The filter is widely used for non-destructive image enhancement. The novelty of the filter is that the processing is performed jointly in the spatial and range (photometric) domain. This, for example, ensures edge preservation during noise cleaning and the enhancement of psycho-visual appearance of an image without the appearance of phantom colors. Unfortunately, the spatio-range processing renders the filter non-linear and computation intensive. In this paper, we present a concise review of the various fast algorithms proposed by researchers for speeding up the filtering without compromising the visual quality. In particular, we explain how these proposals can be clubbed under various broad categories from an approximation-theoretic perspective. We also present simulation results to compare the accuracy-timing trade-offs achieved by some of the state-of-the-art methods