Excellent smoothing algorithm should be able to deal with images corrupted by noise and has the capability of preserving detailed information. Robust filtering is a very easy algorithm

but it has disadvantages that it cannot filter images which are highly corrupted. In this paper

both the traditional robust filtering and its existing improved algorithm are analyzed and realized. Meanwhile

a new adaptive robust filtering is proposed. Based on platform MATALB

comparing the proposed algorithm and these existing algorithms proved that this new algorithm can work properly when images are highly corrupted and preserve image details better than those present algorithms.