Benchmarking Screen Content Image Quality Evaluation in Spatial Psychovisual Modulation Display System

Spatial Psychovisual Modulation (SPVM) is a novel information display technology which aims to simultaneously generate multiple visual percepts for different viewers on a single display. The SPVM system plays an important role in information security. In a SPVM system, the viewers wearing polarized glasses can see a specific image (called personal view), and meanwhile the viewers not wearing glasses can also see a semantically meaningful image (called shared view). Researches on screen content image (SCI) are very hot recently, which have received a great amount of attention from multiple fields in multimedia signal processing. In this paper, we focus our gaze on how the users’ quality-of-experience on SCIs is influenced under the SPVM display system. To this aim, we implement a comprehensive subjective quality assessment of SCIs by building a database which contains the distorted SCIs generated by a SPVM system. We run prevailing image quality methods on the newly established database, and results of experiments indicate that existing image quality metrics cannot reach a good performance, possibly due to some unique distortions, e.g. false contour and ghosting artifacts of SPVM-generated SCIs. Furthermore, we also point out some potential features which may lead to a high-performance metric by some appropriate modification and combination.