High Density Salt and Pepper Noise Removal by a Threshold Level Decision based Mean Filter

An effective algorithm in the switching filter category to remove salt and pepper noise from images at low, as well as high noise density, is presented in this paper. Based on the intensity of pixels of an image, pixels are divided into two classes, noise-free pixel and noisy pixel. The noise-free pixels are left unprocessed and noisy pixel are undergone a filtering process. A threshold value of noise density of the entire image is compared with the threshold level of dimension of the image and depending on it the noisy pixel is processed by mean of the noise-free pixel. The proposed algorithm operates on the boundary pixels and contributes to the edge detection effectively. The simulation is done on Lena and Cameraman image and the tested parameters are the Peak Signal to Noise Ratio (PSNR) and Image Enhancement factor (IEF). PSNR at high noise density (>80%) is greater than any other mentioned filter.