Adaptive working window for impulse noise reduction

We propose an effective method to choose an adaptive working window for impulse noise reduction. Every pixel has the pixel-correlation between itself and its neighbor pixels in an image. Based on the pixel-correlation, we define a formula to gather the statistics for impulse noise intensity estimation. Because the image corrupted by impulse noise would degrade the pixel-correlation, a simple rule is introduced for different median-based filters which are developed with adaptive working window to reduce impulse noise. The reported multi-state median (MSM) filter and switching median (SM) filter are simulated with the simple rule. The experimental results show that our method would improve the reliability of MSM filter and SM filter for impulse noise reduction.