Robust median filter with adaptive window length

The authors introduce a novel adaptive median filter algorithm which efficiently exploits the benefits of the variable size data window. The method exploits the noise rejecting properties of the long window sizes and the edge preservation of the short windows. Adaptive window lengths reduce the variance of the filtered signal considerably without degrading the accurate representation of the discontinuities. Finally, a computationally efficient implementation of the proposed adaptive median filter is presented. Examples show that the proposed method outperforms the conventional fixed window length median operators