An optical flow direction-based full reference video quality assessment algorithm. (12th January 2021)

• Record Type: Journal Article
• Authors: Gujjunoori, Sagar; Oruganti, Madhu
• Is Part Of: International journal of high performance computing and networking. Volume 16:Number 2/3(2020)
• Page Start: 148
• Page End: 159
• Publication Date: 2021-01-12
• Subjects: optical flow -- direction -- video quality assessment -- VQA -- human visual system -- HVS; High performance computing -- Periodicals; Computer networks -- Periodicals; High performance computing; Periodicals; 004.05
• Languages: English
• ISSNs: 1740-0562

Abstract

Modelling of functions of visual cortex and optical flow field has made it possible to develop many sophisticated human visual system-based visual quality assessment algorithms. There exist very few video quality assessment algorithms which consider spatio-temporal distortions effectively. In this article, based on the direction of optical flow field, we propose a full reference video quality algorithm. We claim that some temporal distortions, which are present in other directions than the majority of optical flow field directions, can not be easily perceivable by the human observer; hence it results in reduced video quality scores. Our proposed method achieved high video quality score and higher correlation with formally existed algorithm and near to the objective scores. Our experimental results demonstrate that the proposed algorithm improves the state of art.