Spectral image distortion map

Abstract

In this paper, a novel technique of spectral image quality evaluation using spectral image distortion map (SIDM) is proposed. The method is based on a recent approach to evaluation of color differences in a spectral space. What is calculated here, in fact, is a pixelwise spectral distortion. As the measure of the dissimilarity, a novel kernel based similarity measure is used. The metric produces comparable values of differences for perceptually equally disparate colors. As a result, a grayscale spectral distortion image is obtained, where the intensity of each of the pixels is a difference between the original image and the distorted one. A perceptual image distortion map (PIDM) has also been constructed to show the accuracy of SIDM. A comparison of PIDM and SIDM shows that the latter provides an excellent fit to the response of the human visual system.