• DocumentCode
    614544
  • Title

    Objective assessment of surveillance video quality

  • Author

    Petroviu, Vladimir ; Bondzulic, Boban

  • Author_Institution
    Imaging Sci. & Biomed. Eng., Univ. of Manchester, Manchester, UK
  • fYear
    2012
  • fDate
    25-27 Sept. 2012
  • Firstpage
    1
  • Lastpage
    5
  • Abstract
    This paper presents an objective video quality index capable of subjectively relevant video quality and signal fidelity assessment. Surveillance and security video is often transmitted through constrained channels leading to serious, mainly compression borne degradations of received signals. The value of such streams to an observer tasked with viewing them is directly determined by their fidelity and perceived quality. Our video quality index evaluates preservation of localised gradient and colour information from the original reference video in the transmitted, degraded version as a basis of an overall fidelity score. Applying a biologically inspired information loss model yields a perceptually relevant quality score for each degraded sequence. Our approach is shown to be able to highlight areas where true scene information is lost and compression artifacts appear as well as provide quality scores with a very high level of agreement with subjective assessments obtained through time consuming psychometric trials on a wide range of surveillance video quality.
  • Keywords
    gradient methods; image colour analysis; image sequences; natural scenes; video coding; video surveillance; biologically inspired information loss model; colour information preservation; compression artifacts; compression borne received signal degradation; degraded sequence; fidelity score; localised gradient information preservation; objective assessment; objective video quality index; perceived quality; quality score; security video; signal fidelity assessment; surveillance video quality; true scene information loss;
  • fLanguage
    English
  • Publisher
    iet
  • Conference_Titel
    Sensor Signal Processing for Defence (SSPD 2012)
  • Conference_Location
    London
  • Electronic_ISBN
    978-1-84919-712-0
  • Type

    conf

  • DOI
    10.1049/ic.2012.0105
  • Filename
    6552173