Digital image filtering: An adaptive median filter.

摘要

In this thesis, a new adaptive median filter (AMF) is presented which combines the median filter (MF) and the multilevel median filter (MMF) to smooth impulsive noise. The window size of the MF can be adjusted according to the width of noise, with the window size of the MMF fixed. The proposed adaptive median filter switches to the MF or the MMF automatically according to local characteristics present in the image. Experimental results are presented to demonstrate the performance of the proposed adaptive median filter.