Efficient Full-Reference Assessment of Image and Video Quality

Automatic quality assessment of digital pictures is a crucial issue in several image and video processing applications including broadcasting, archiving, or restoration. The visibility of impairments related to digital image processing systems is subject to the spatio-temporal properties of the given image or video content. As quality assessment using subjective tests carried out by humans is very costly, time consuming, and not compatible with real-time constraints, objective measures are required that can predict the perceptual judgment of human viewers. In this paper, a top-down quality assessment tool that mimics a selection of prominent human visual system properties is presented. Quality evaluation is conducted based on perceptually salient feature points in our approach. It is shown that the proposed method, although featuring a significantly lower complexity than standardized quality measures, performs as well as these for block-based hybrid video coding.