An Enhanced Impulse Noise Filter Using the Homogeneity Level of the Local Region

This paper proposes a new way to detect whether the current pixel is contaminated, which is based on the homogeneity level of the local region around that pixel. Then, we combine this idea with the Adaptive impulse detection using center-weighted median (ACWM) filter and Directional weighted median (DWM) filter to get the new efficient filter: Region Filter (RF). Extensive experiments show that with the low noise level (≤ 25%), the RF not only outperforms many well-known techniques, but also can preserve highly detail features.