Impulse Noise Restoration Using Combined Fuzzy Logic and Adaptive Trimmed Median Filter

This work focuses on the development of algorithm for the removal of impulse noise from digital images in spatial domain in low density as well as high density of noises. Various issues related to the noisy problem are studied for grayscale as well as color images and suitable filtering methods are suggested. In this paper, fuzzy logic is used to detect the noisy pixels in an impulse noise corrupted image and an adaptive trimmed median filtering technique is proposed to remove the detected noisy pixel from gray and color images. When some of the pixels in a considered window of an image are noisy pixels, then median value of the noise-free pixels in that window replaces the processing detected noisy pixel. However, when all the pixels in a selected window are all noisy pixels, then in such cases the possible solution is to replace the processing pixel by the mean and standard deviation values of the elements in the selected window. The combined fuzzy logic and adaptive trimmed median filter approach is also used to preserve the edges and fine details of the images. To assess the performance of the proposed method, several standard grayscale and color test images are used in the experiments which have distinctly different features. The efficacy of the various filtering systems is evaluated both qualitatively and quantitatively in terms of PSNR and MSE for grayscale as well as color images.