Psycho-visual based distortion measures for monochrome image and video compression

Describes quantitative distortion measures for compressed monochrome image and video based on a psycho-visual model. The model follows the human vision perception in that the distortion as perceived by a human viewer is dominated by the compression error uncorrelated with the local features of the original image and for a video sequence the distortion is perceived from two sources, the still areas and the motion areas of a video frame. The authors have performed subjective tests to obtain ranking results for compressed images and video sequences which were compressed using different compression algorithms and compared the results with the rankings obtained using their distortion measure and other existing mean-square error based measures. They have found that their distortion measure´s ranking matches the subjective ranking accurately whereas the mean-square error and its variants performed poorly in matching the subjective ranking on an average