An Innovative Video Quality Assessment Method and An Impairment Video Dataset

Hang Shi, Chengjun Liu

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

1 Scopus citations

Abstract

Impairments in video adversely affect the performance of computer vision applications. This paper presents a comprehensive impairment video dataset and proposes an innovative video quality assessment (VQA) method to evaluate the video impairment levels. First, a dataset of 300 short videos (see www.cdvl.org) is developed with representative impairments: 19 types of individual impairments, and 35 types of combined impairments (see www.iaitusa.com for the specific types of impairments). Second, an innovative no-reference (iNR) metric vector is presented to assess the video impairment levels. In particular, the iNR metric vector is defined by five impairment scores: The blur impairment score, the small noise patches impairment score, the whole frame illumination impairment score, the partial frame illumination impairment score, and the temporal impairment score. The iNR metric vector thus is not only able to define a novel failure rate (FR) for each video to characterize the nature of the leading impairment, but it can also quantify the impact caused by other impairments in the same video.

Original languageEnglish (US)
Title of host publicationIST 2021 - IEEE International Conference on Imaging Systems and Techniques, Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781728173719
DOIs
StatePublished - 2021
Event2021 IEEE International Conference on Imaging Systems and Techniques, IST 2021 - Virtual, New York, United States
Duration: Aug 24 2021Aug 26 2021

Publication series

NameIST 2021 - IEEE International Conference on Imaging Systems and Techniques, Proceedings

Conference

Conference2021 IEEE International Conference on Imaging Systems and Techniques, IST 2021
Country/TerritoryUnited States
CityVirtual, New York
Period8/24/218/26/21

All Science Journal Classification (ASJC) codes

  • Artificial Intelligence
  • Computer Science Applications
  • Computer Vision and Pattern Recognition
  • Signal Processing
  • Decision Sciences (miscellaneous)

Keywords

  • impairment video dataset
  • innovative no-reference (iNR) metric vector
  • novel failure rate (FR)
  • video quality assessment (VQA)

Fingerprint

Dive into the research topics of 'An Innovative Video Quality Assessment Method and An Impairment Video Dataset'. Together they form a unique fingerprint.

Cite this