WAF and WMF to improve the performance of average and median filters

Abstract

Salt and spice noise is considered one of the most common types of noise that affect grayscale and color digital images, as it affects them negatively, and this negative effect increases with an increase in the noise ratio. Many digital filters are used to mitigate the negative effects of salt and pepper noise, and the most widely used of these filters are the average filter and the median filter. The average and median filters work on processing all pixels in the image, whether these pixels are intact from the noise or infected with it, and accordingly, mitigating this noise, especially if it has a high noise ratio, is ineffective. In this research paper, new window average and window median filters will be proposed to enhance the performance of standard average and median filters. The proposed filter will treat the infected pixels, leaving the clean pixels as they were. For a noisy pixel, a special window will be created and the pixel value will equal the mean (or average) of the cleaned pixels in the window (excluding the noisy pixels). To simplify the window processing an index window will be used, this window will point to the noisy and cleaned pixels in the selected pixel window. Several images with various values of noise ratios will be tested; several windows with different sizes will be examined to get the most suitable window size. The selected window size will be used to filter various noisy images, the obtained results will be compared with average and median filter results to show the improvements provided by the proposed method.