Publication detail

Concept of Variation Detector Used in Video Signal Compression Domain

FRÝZA, T. HANUS, S.

Original Title

Concept of Variation Detector Used in Video Signal Compression Domain

Czech Title

Princip detekce střihu ve video sekvenci v oblasti komprese video signálů

English Title

Concept of Variation Detector Used in Video Signal Compression Domain

Type

journal article

Language

en

Original Abstract

This paper is concerned with a proportion similarity of the substance in images. In particular, we investigate the possibility of finding a cut position in a video sequence. The cut position detector is based on the data comparison when for two successive frames of the video sequence, several parameters are being evaluated. In this contribution we deal only with parameters in time domain, such as mean square and mean absolute error, image fidelity, normalized cross-correlation, Czekanowski distance and normalized distance. All functions are tested with a video sequence containing two cuts. Finally, a qualitative comparison of proposed methods is performed. For the cut detection in time domain is sufficient to use only two parameters: mean absolute error and normalized distance. Besides, for the color input data, the detector may scan only the luminance signal Y of YCbCr representation.

Czech abstract

Článek se zabývá podobností jednotlivých snímků ve video sekvenci. Konkrétně popisuje možnost detekce střihu ve video sekvenci pro potřeby optimalizace 3D DCT transformace.

English abstract

This paper is concerned with a proportion similarity of the substance in images. In particular, we investigate the possibility of finding a cut position in a video sequence. The cut position detector is based on the data comparison when for two successive frames of the video sequence, several parameters are being evaluated. In this contribution we deal only with parameters in time domain, such as mean square and mean absolute error, image fidelity, normalized cross-correlation, Czekanowski distance and normalized distance. All functions are tested with a video sequence containing two cuts. Finally, a qualitative comparison of proposed methods is performed. For the cut detection in time domain is sufficient to use only two parameters: mean absolute error and normalized distance. Besides, for the color input data, the detector may scan only the luminance signal Y of YCbCr representation.

Keywords

Video signal coding, cut detection, interframe coding, 3D DCT, video cube, transparency

RIV year

2004

Released

26.10.2004

Pages from

1776

Pages to

1780

Pages count

5

BibTex


@article{BUT42165,
  author="Tomáš {Frýza} and Stanislav {Hanus}",
  title="Concept of Variation Detector Used in Video Signal Compression Domain",
  annote="This paper is concerned with a proportion similarity of the substance in images. In particular, we investigate the possibility of finding a cut position in a video sequence. The cut position detector is based on the data comparison when for two successive frames of the video sequence, several parameters are being evaluated. In this contribution we deal only with parameters in time domain, such as mean square and mean absolute error, image fidelity, normalized cross-correlation, Czekanowski distance and normalized distance. All functions are tested with a video sequence containing two cuts. Finally, a qualitative comparison of proposed methods is performed. For the cut detection in time domain is sufficient to use only two parameters: mean absolute error and normalized distance. Besides, for the color input data, the detector may scan only the luminance signal Y of YCbCr representation.",
  chapter="42165",
  journal="WSEAS Transactions on Circuits",
  number="9",
  volume="3",
  year="2004",
  month="october",
  pages="1776",
  type="journal article"
}