An Innovative No-Reference Metric for Real-Time 3D Stereoscopic Video Quality Assessment

Yi Han, Zhenhui Yuan, Gabriel-Miro Muntean

Research output: Contribution to journalArticlepeer-review

20 Citations (Scopus)

Abstract

3D video technologies have been widely adopted by video service providers and consumer electronics stakeholders due to their potential of offering an immersive user experience. In case of 3D video streaming, the dynamic network conditions are the bottleneck that limits the content delivery at good perceived quality levels and an effective solution is to employ advanced 3D video adaptation schemes. Accurate real-time objective 3D video quality assessment is a critical factor in adaptive decision making. State-of-the-art objective 3D video quality assessment methods are in general reference-based and require the availability of the original 3D video sequence, which makes them not suitable for real-time applications. This paper proposes the extended no reference objective video quality metric (eNVQM), an innovative metric for real-time 3D video quality assessment. eNVQM estimates the 3D video quality by taking as the input parameters network packet loss, video transmission bitrate, and frame rate. Based on extensive subjective tests, eNVQM models the impact of network packet loss on 3D video at different bitrates and frame rates on the perceived stereoscopic 3D video quality. The performance of eNVQM is investigated by comparing its results with two state-of-the-art objective video quality metrics: 1) structural similarity index and 2) video quality metric. Results show that eNVQM maintains similar accuracy level in estimating 3D video quality with the alternative reference-based metrics.
Original languageEnglish
Pages (from-to)654-663
Number of pages10
JournalIEEE Transactions on Broadcasting
Volume62
Issue number3
Early online date30 Mar 2016
DOIs
Publication statusPublished - Sept 2016
Externally publishedYes

Keywords

  • 3D video
  • objective quality assessment
  • non-reference
  • stereoscopic

Fingerprint

Dive into the research topics of 'An Innovative No-Reference Metric for Real-Time 3D Stereoscopic Video Quality Assessment'. Together they form a unique fingerprint.

Cite this