Perceptual quality measurement of video frames affected by both packet losses and coding artifacts

Tao Liu, Hua Yang, Alan Stein, Yao Wang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This paper addresses the problem of perceptual quality measurement on individual coded video frames with distortions caused by both lossy source coding and lossy transmission. We focus on low bitrate H.264 coded video with error concealment and propagation due to packet losses. Identifying Human Visual System (HVS) masking effect as one of the most important factors affecting the perceptual quality, we investigate its impacts on the two aforementioned artifacts in individual video frames respectively, and then evaluate their corresponding Just-Noticeable-Difference (JND) based perceptual distortions. Finally, we combine the two distortions together and propose a block-wise full-reference quality metric for a single distorted video frame. Our subjective evaluation results show that the proposed metric correlates fairly well with actual subjective quality ratings.

Original languageEnglish (US)
Title of host publication2009 International Workshop on Quality of Multimedia Experience, QoMEx 2009
Pages210-215
Number of pages6
DOIs
StatePublished - 2009
Event2009 International Workshop on Quality of Multimedia Experience, QoMEx 2009 - San Diego, CA, United States
Duration: Jul 29 2009Jul 31 2009

Publication series

Name2009 International Workshop on Quality of Multimedia Experience, QoMEx 2009

Other

Other2009 International Workshop on Quality of Multimedia Experience, QoMEx 2009
Country/TerritoryUnited States
CitySan Diego, CA
Period7/29/097/31/09

Keywords

  • Error propagation
  • H.264/AVC
  • Just noticeable difference
  • Perceptual quality
  • Spatial masking

ASJC Scopus subject areas

  • Computer Graphics and Computer-Aided Design
  • Computer Science Applications
  • Software

Fingerprint

Dive into the research topics of 'Perceptual quality measurement of video frames affected by both packet losses and coding artifacts'. Together they form a unique fingerprint.

Cite this