Full reference and reduced reference metrics for image quality assessment

In this paper, we propose efficient full reference and reduced reference image quality assessment metrics. These two metrics are based on human visual system properties to get the best correspondence with human judgments. The full reference metric is generic (independent from image distortion type) and convenient for image coding schemes comparison. From this metric, we have derived a reduced reference metric suitable for quality of service monitoring in a broadcasting purpose. The two metrics share common processings. The main difference between them comes from the selection of information in order to get a high-level representation of images for a reduced reference metric. If limited to a specific type of image coding scheme, the two metrics achieve similar performances.