Window mode filter for image denoising

We formulate the window mode filter (WMF) which is an extension of the point-wise mode filter to windowed version. We then derive from the WMF the window-weighted bilateral filter (WWBF) which is different from the previously presented window bilateral filter (WBF). We demonstrate with experiments that the WWBF outperforms WBF for images contaminated with Gaussian noises and those with mixed Gaussian and impulsive noises. We also show that the PSNR of iterated WWBF is always larger than the iterated WBF.