Adaptive decimated median filters for image processing

The use of 2-D decimated median filters (DMFs) for the removal of impulsive noise and the preservation of edges is proposed. The 2-D DMF preserves edges and corners better than the median filter because its root signal set contains corners as well as edges. Its impulse noise removal capability is comparable with that of the median filter while its performance near edges is superior. The computational complexity of the DMF is lower than that of the median filter. The authors investigate the deterministic and statistical properties of the DMF and show that it performs best with a particular decimation scheme. Based on this observation, a way of adapting the decimation with respect to edge orientation is proposed. This filter is used to remove noise from a test image that contains horizontal, vertical, and diagonal edges.<>