Contrast Weighted Perceptual Structural Similarity Index for Image Quality Assessment

In this paper a full reference objective image quality assessment technique is presented which is based on the properties of the human visual system (HVS). By integrating the notion of perceptually important regions with the measurement of structural similarity between the original image and distorted image a contrast weighted Perceptual Structural SIMilarity Index PSSIM^c is proposed. The method first evaluates the structural similarity indices between the original and distorted image in local regions. These local indices are then weighted based on the perceptual weights of the corresponding region, characterized by the contrast value of the local region. PSSIM^c of an image is calculated as the average of these weighted indices. A comparison with the peak-signal-to-noise ratio (PSNR) and state of the art metric, Mean Structural Similarity Index (MSSIM), shows that the proposed measure correlates better with the judgment of human observers.