Mots-clé
(Anglais)
|
Quality of experience; 3DTV; stereoscopic video; multi view video; subjective quality methods; objective quality metrics; motion parallax; asymmetric stereo; quality metric; paired comparison
|
Programme de recherche
(Anglais)
|
COST-Action IC1003 - European Network on Quality of Experience in Multimedia Systems and Services (QUALINET)
|
Description succincte
(Anglais)
|
From all the emerging multimedia modalities, 3D television (3DTV) is currently by far the one with the largest momentum, due to the availability of interesting 3D content and affordable 3D display technologies. However, one of the problems that may prevent 3DTV from becoming a world-wide success is that users expect an improved experience with respect to conventional TV and HDTV. Unfortunately, current 3DTV does not always offer a better Quality of Experience (QoE). In some situations, the perceived quality may become lower for the same content shown in 3D when compared to 2D. Reasons for that include technological limitations of current 3D technologies and lack of knowledge for 3D content creation. In order to ensure the adoption of 3DTV in the next decade, it is therefore crucial to optimize the 3D quality from a user perspective along the whole 3D processing chain. However, this requires reliable tools for 3D quality evaluation. While 2D quality evaluation is quite mature, 3D quality evaluation is a rather unexplored topic. The goal of this project is to advance the state of the art in 3D video quality evaluation by developing more reliable subjective quality methods and objective quality metrics that consider quality from a user perspective. This will be achieved by analyzing the requirements of a home entertainment scenario and creating a representative dataset which considers not only the content, but also the context and the users. This dataset will be used to compare existing approaches for 3D quality evaluation and identify their strength and weaknesses, based on which more reliable methods will be developed. With these objectives the project fits very well into the recently approved COST Action IC1003, which creates network of experts working on Quality of Experience in multimedia systems and services.
|
Partenaires et organisations internationales
(Anglais)
|
AT, BE, CH, CY, CZ, DE, DK, EL, ES, FI, FR, HR, IE, IT, MT, NO, PL, PT, RS, SE, SI, SK, UK
|
Résumé des résultats (Abstract)
(Anglais)
|
Quality is a fundamental concept closely linked to measuring performance of systems, products and services. Any compression and representation technique in audiovisual signals needs metrics of quality, not only in their design stage but also in demonstrating their superior performance when compared to state of the art. Some believe that the main reason for the failure of previous waves of 3D was the lack of a minimum quality requirement in 3D. It is therefore important to consider mechanisms of 3D vision in humans, and their underlying perceptual models, in conjunction with the type of distortions that today and tomorrow 3D video processing systems produce. In particular, there is a need to understand how concepts known in 2D can be extended to 3D, and where they need to be entirely reinvented from scratch, in order to lead to more efficient methods for assessing subjective and objective quality of 3D visual content. Despite the efforts of the scientific community in recent years, 3D video quality assessment is still an open challenge and there are no metrics which are widely recognized as reliable predictors of human 3D quality perception. The assessment of 3D quality is particularly challenging for 3D video since many artifacts can alter the 3D content, especially when the rendered stereo pair is formed by two views having different visual quality (mismatched or asymmetric stereo pair). In this case, specific properties of the human visual system, such as binocular suppression (the masking of low-frequency content in one view by the sharp visual content in the other view), should be taken into account when building models that objectively quantify the overall quality of the rendered stereo pair.
|
Références bases de données
(Anglais)
|
Swiss Database: COST-DB of the State Secretariat for Education and Research Hallwylstrasse 4 CH-3003 Berne, Switzerland Tel. +41 31 322 74 82 Swiss Project-Number: C10.0132
|
