A Decision Based Asymmetrically Trimmed Modified Geometric Mean Algorithm for the Removal of High Density Salt and Pepper Noise in Images and Videos

Abstract

A decision based asymmetrically trimmed modified geometric mean algorithm (DBATMGMA) for the removal of high density salt and pepper noise in images and videos is proposed. The algorithm uses fixed 3 × 3 windows for increasing noise densities. The algorithm initially checks for the presence of outliers (0 or 255) in the processed pixel. If it holds the outliers, the processed pixel is termed faulty. If the processed pixel is faulty, check for the four neighbors are noisy or not (0 or 255). If all the four neighbors are noisy, then check for all the pixels is noisy in the confined neighborhood. If the entire window is noisy, then the corrupted pixel is replaced by mean of all the elements else the corrupted pixel is replaced by mean of the four neighbors. If all the four neighbors are not noisy, then the faulty pixels are replaced with asymmetrically trimmed modified geometric mean. If the pixels does not hold the outlier, then the pixel is considered undamaged and left unaltered. The proposed algorithm is compared with standard algorithms on an image database. The noise suppression along with information preservation capability of the proposed algorithm was found to be very good both in terms of qualitative and quantitative measures.