Information Content Redundancies of Image Quality Assessments
Jens Preiss, Philipp Urban · 2011
In this paper we investigate the comparison of image quality assessment (IQA) methods for color images. An IQA evaluates a pair of images and returns a single number as a prediction of the perceived image difference. We introduce the Redundant Information Content Evaluation (RICE) that detects redundancies between distinct IQAs. Such comparisons help to develop or improve IQAs with respect to the information that they extract from the images. RICE is analyzed using a set of common color IQAs applied on a large image difference database. Its functionality and performance is tested and an application for color image difference analysis is proposed.