High performance decision based median filter for salt and pepper noise removal in images

Abstract

A high performance decision based median filter is proposed for removal of salt and pepper in images. It is an enhanced Adaptive Switching Median filter which initially detects noise pixels iteratively through several phases and replaces the noisy pixels with median value. It calculates median value without considering noisy pixels to improve the performance of median filter for high density noise. Detection of noise is done by expanding the mask until 7×7 to maintain local information extraction. Moreover, the processing pixel is replaced by last processed pixel if the algorithm fails to detect noise free pixel at 7×7. If the noise free median value is not available at 7×7 processing window, the last processed pixel take into consideration if it is noise free. If the last processed pixel is noisy, the algorithm select a window size with 15×15 dimension and calculate the number of 0's and 255's in the processing window. Then replace the processing pixel with 0 or 255 which is more in number in the selected window. Experiment result shows that it can provide very high quality restored images for images that are contaminated by “salt & pepper” noise, especially when the noise density is large.