Extracting objects and events from MPEG sequences for video highlights indexing and retrieval

Jinchang Ren, J. Jiang , J. Chen, S. Ipson

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

Automatic recognition of highlights from videos is a fundamental and challenging problem for content-based indexing and retrieval applications. In this paper, we propose techniques to solve this problem using knowledge supported extraction of semantics, and compressed-domain processing is employed for efficiency. Firstly, knowledgebased rules are utilized for shot detection on extracted DCimages, and statistical skin detection is applied for human object detection. Secondly, through filtering outliers in motion vectors, improved detection of camera motions like zooming, panning and tilting are achieved. Video highlight high-level semantics are then automatically extracted via low-level analysis in the detection of human objects and camera motion events, and finally these highlights are taken for shot-level annotation, indexing and retrieval. Results using a large test video data set have demonstrated the accuracy and robustness of the proposed techniques.
LanguageEnglish
Pages95-103
Number of pages9
JournalJournal of Multimedia
Volume5
Issue number2
DOIs
Publication statusPublished - May 2010

Fingerprint

Semantics
Cameras
Skin
Processing
Object detection

Keywords

  • content-based retrieval
  • compressed domain processing
  • video semantics
  • video highlights extraction

Cite this

@article{f838962860644b67b61b85eaa2dd2c69,
title = "Extracting objects and events from MPEG sequences for video highlights indexing and retrieval",
abstract = "Automatic recognition of highlights from videos is a fundamental and challenging problem for content-based indexing and retrieval applications. In this paper, we propose techniques to solve this problem using knowledge supported extraction of semantics, and compressed-domain processing is employed for efficiency. Firstly, knowledgebased rules are utilized for shot detection on extracted DCimages, and statistical skin detection is applied for human object detection. Secondly, through filtering outliers in motion vectors, improved detection of camera motions like zooming, panning and tilting are achieved. Video highlight high-level semantics are then automatically extracted via low-level analysis in the detection of human objects and camera motion events, and finally these highlights are taken for shot-level annotation, indexing and retrieval. Results using a large test video data set have demonstrated the accuracy and robustness of the proposed techniques.",
keywords = "content-based retrieval , compressed domain processing, video semantics, video highlights extraction",
author = "Jinchang Ren and J. Jiang and J. Chen and S. Ipson",
year = "2010",
month = "5",
doi = "10.4304/jmm.5.2.95-103",
language = "English",
volume = "5",
pages = "95--103",
journal = "Journal of Multimedia",
issn = "1796-2048",
publisher = "Academy Publisher",
number = "2",

}

Extracting objects and events from MPEG sequences for video highlights indexing and retrieval. / Ren, Jinchang; Jiang , J.; Chen, J.; Ipson, S.

In: Journal of Multimedia, Vol. 5, No. 2, 05.2010, p. 95-103.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Extracting objects and events from MPEG sequences for video highlights indexing and retrieval

AU - Ren, Jinchang

AU - Jiang , J.

AU - Chen, J.

AU - Ipson, S.

PY - 2010/5

Y1 - 2010/5

N2 - Automatic recognition of highlights from videos is a fundamental and challenging problem for content-based indexing and retrieval applications. In this paper, we propose techniques to solve this problem using knowledge supported extraction of semantics, and compressed-domain processing is employed for efficiency. Firstly, knowledgebased rules are utilized for shot detection on extracted DCimages, and statistical skin detection is applied for human object detection. Secondly, through filtering outliers in motion vectors, improved detection of camera motions like zooming, panning and tilting are achieved. Video highlight high-level semantics are then automatically extracted via low-level analysis in the detection of human objects and camera motion events, and finally these highlights are taken for shot-level annotation, indexing and retrieval. Results using a large test video data set have demonstrated the accuracy and robustness of the proposed techniques.

AB - Automatic recognition of highlights from videos is a fundamental and challenging problem for content-based indexing and retrieval applications. In this paper, we propose techniques to solve this problem using knowledge supported extraction of semantics, and compressed-domain processing is employed for efficiency. Firstly, knowledgebased rules are utilized for shot detection on extracted DCimages, and statistical skin detection is applied for human object detection. Secondly, through filtering outliers in motion vectors, improved detection of camera motions like zooming, panning and tilting are achieved. Video highlight high-level semantics are then automatically extracted via low-level analysis in the detection of human objects and camera motion events, and finally these highlights are taken for shot-level annotation, indexing and retrieval. Results using a large test video data set have demonstrated the accuracy and robustness of the proposed techniques.

KW - content-based retrieval

KW - compressed domain processing

KW - video semantics

KW - video highlights extraction

U2 - 10.4304/jmm.5.2.95-103

DO - 10.4304/jmm.5.2.95-103

M3 - Article

VL - 5

SP - 95

EP - 103

JO - Journal of Multimedia

T2 - Journal of Multimedia

JF - Journal of Multimedia

SN - 1796-2048

IS - 2

ER -