Study on fast algorithm of median filtering based on DC and ICS method

摘要

This paper studies median filtering algorithm and several related fast algorithm proposed by other professors, then gives a novel fast algorithm of median filtering based on divide-and-conquer method and improved-choose-sort method. In new algorithm, all the pixels in filtering window are divided into several blocks, the median values of each block are computed using improved-choose-sort algorithm, then consider the above median values as a new block and compute its median value again, which can be seen as the median value of all the pixel values. In ICS algorithm, only half of the block data are sorted and the rest can be ignored. Experiment results show that the new algorithm is significantly superior to classical algorithm especially in large window cases; it can greatly improve the performance of median filtering algorithm while keeping very high accuracy.