Full reference-image quality assessment with global and local image regions for visual quality perception and optimization

Abstract

Image Quality Assessment (IQA) by using mathematical methods is offering favorable results in calculating visual quality of distorted images. These methods are developed by examining effective features that are compatible with characteristics of Human Visual System (HVS). But many of those methods are difficult to apply for optimization problems. This paper presents DCT based metric with easy implementation and having mathematical properties like differentiability, convexity and valid distance metricability to overcome the optimization problems. By using this method we are able to calculate the quality of image as a whole as well as the quality of local image regions.