High Density Impulse Noise Removal Based on the Total Observation Kernel Element for Image Sequences

Abstract

Several different methods for impulse noise removal in image sequences have been proposed. However, all of them are not successful in removing high density of impulse noise. Hence, this paper proposes a filtering method for reducing high density impulse noise in the image sequences. We use three windows with size 3 × 3 to obtain a new window with similar size. Three windows are taken from the next-frame, current frames and previous frames. The recursive window is applied in the current frames. The filtering process uses decision-based method. Meanwhile, a pixel for replacing the noisy pixel is calculated from a new window based on weighting method. Our experimental results show that the proposed method can not only reduce the high impulse noise in image sequences well, but also preserve more details and textures.