Recursive bilateral filtering (developed by Qingxiong Yang) is pretty fast compared with most edge-preserving filtering methods

• computational complexity is linear in both input size and dimensionality:
• takes about 43 ms to process a one megapixel color image (i7 1.8GHz & 4GB mem)
• about 18x faster than Fast high-dimensional filtering using the permutohedral lattice
• about 86x faster than Gaussian kd-trees for fast high-dimensional filtering

For more details of the algorithm, please refer to the original paper

@inproceedings{yang2012recursive,
 title={Recursive bilateral filtering},
 author={Yang, Qingxiong},
 booktitle={European Conference on Computer Vision},
 pages={399--413},
 year={2012},
 organization={Springer}
}