High density impulse noise reduction by denoising neighbor pixels

Abstract

A new method based on adaptive median filtering is proposed in this paper to remove high density impulse noise. This method replaces central noisy pixel having gray scale value 0 or 255 by median value of 3×3 window. Algorithm is proposed to replace the noisy neighbor pixels in the window. The clean pixels are then used to calculate the median value, which then replaces noisy central pixel. Proposed method has been tested on images which were contaminated by impulse noise of different densities, and results are compared with existing simple median filter with window size of 3 × 3, 5 × 5, and adaptive median filter. Results shows that Proposed method gives better Mean Square Error (MSE), Universal Quality Index (UQI), and Peak Signal to Noise Ratio (PSNR), when compared with the simple and adaptive median filter.