After analyzing several schemes for determining filtering window size for the removal of salt-and-pepper noise

we propose a distance-weighted image denoising method based on adaptive filtering window. The proposed method first identifies the pixels with gray level 0 or 255 as noise pixels. Then

for each noise pixel

the minimum window with noise-free pixels is found. If the minimum window of a certain noise pixel is smaller than a given threshold

then noise-free pixels within the minimum window are used to perform distance-weighted filtering. Otherwise

the current noise pixel should be located in the regions composed of noise-free pixels with gray level 0 or 255

and a majority strategy is used to generate the restored gray level. The proposed method is evaluated by comparing it with seven other image denoising methods. Simulation results show that the new method achieves a better effect than its counterparts on images that contain noise-free pixels with gray level 0 or 255. Among all the compared methods

the new method achieves the best denoising effect on images that contain few or even no pixels with gray level 0 or 255. The proposed method effectively removes salt-and-pepper noise and is also suitable for images that contain many noise-free pixels with gray level 0 or 255.