Abstract

An image degraded by noise is a common phenomenon. In this letter, we propose a novel adaptive fuzzy switching weighted mean filter to remove salt-and-pepper (SAP) noise. The process of denoising includes two stages: noise detection and noise elimination. In the first stage, pixels in a corrupted image are classified into two categories: original pixels and possible noise pixels. For the latter, we compute the maximum absolute luminance difference of processed pixels next to possible noise pixels to classify them into three categories: uncorrupted pixels, lightly corrupted pixels, and heavily corrupted pixels. In the second stage, under the assumption that pixels at a short distance tend to have similar values, the distance relevant weighted mean of the original pixels in the neighborhood of a noise pixel are computed. For a non-noise pixel, retain it as unchanged; for a lightly corrupted pixel, replace it with the weighted average value of the weighted mean and its own value; and for a heavily corrupted pixel, change it to be the weighted mean. Experimental results show that compared to some state-of-the-art algorithms, our method keeps more texture details and is better at removing SAP noise and depressing artifacts.An image degraded by noise is a common phenomenon. In this letter, we propose a novel adaptive fuzzy switching weighted mean filter to remove salt-and-pepper (SAP) noise. The process of denoising includes two stages: noise detection and noise elimination. In the first stage, pixels in a corrupted image are classified into two categories: original pixels and possible noise pixels. For the latter, we compute the maximum absolute luminance difference of processed pixels next to possible noise pixels to classify them into three categories: uncorrupted pixels, lightly corrupted pixels, and heavily corrupted pixels. In the second stage, under the assumption that pixels at a short distance tend to have similar values, the distance relevant weighted mean of the original pixels in the neighborhood of a noise pixel are computed. For a non-noise pixel, retain it as unchanged; for a lightly corrupted pixel, replace it with the weighted average value of the weighted mean and its own value; and for a heavily corrupted pixel, change it to be the weighted mean. Experimental results show that compared to some state-of-the-art algorithms, our method keeps more texture details and is better at removing SAP noise and depressing artifacts.

Paper Details