Quality of Experience for High Definition Presentations - Case: Digital Cinema

Abstract

AbstractWorld-wide roll out of Digital Cinema (D-Cinema) is pushing the boundaries for Ultra High Definition content to all of us - the users. This poses new challenges on assessing our perception of high quality media and also understanding how quality can impact the business model. We use Quality of Experience (QoE) as the term used to describe the perception of how usable or good the users think the media or services are. In order to understand QoE we explore the Human Visual System (HVS) and discuss the impact of HVS on designing a methodology for measuring subjective quality in D-Cinema as a use case. Following our methodology, we describe our laboratory set up at the NOVA kino - a 440 seat fully digitized screen in full cinema production 24/7. This setup is used for subjective assessment of D-Cinema content, applying a test methodology adopted from existing recommendations. The subjective quality results are then used to design and to validate an objective metric. Our experiments demonstrate that traditional quality metrics cannot be adopted in D-Cinema presentations directly. Instead we propose a HVS-based approach to improve the performance of objective quality metrics, such as PSNR, SSIM, and PHVS, in quality assessment of D-Cinema setups. Finally we conclude by discussing how quality impacts the business model, using empirical data from a pilot D-Cinema roll out in Norway – the NORDIC project. This pilot showed that quality can influence users and create more loyalty and willingness to pay a premium if the high quality is used to enhance their experience. Examples are the widespread reintroduction of 3D as well as alternative content such as music, sports, and business related presentations.KeywordsRoot Mean Square ErrorDiscrete Cosine TransformHuman Visual SystemMean Opinion ScoreImage Quality AssessmentThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.