Skip to main navigation Skip to search Skip to main content

Abnormal crowd motion detection using double sparse representations

  • Jinzhe Jiang
  • , Ye Tao
  • , Wei Zhao
  • , Xianglong Tang*
  • *Corresponding author for this work
  • Harbin Institute of Technology

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Sparse representation method has been well used in image analysis, restoration and recognition, and it has also been introduced to analysis of video crowd movements recent years. To improve its accuracy of detecting abnormal events in crowd videos, a double sparse representation method is proposed. The method has two sparse processes, one of them judges whether the region of interest is normal, the other finds out whether the region is abnormal. The two judgments will be processed by fuzzy integral to obtain a final result for this region. Experiments are proceed in different datasets to validate the advantages of our algorithm. The results show that our method achieves higher accuracy than previous methods which are used for analysis of video crowd movements.

Original languageEnglish
Title of host publicationIntelligence Science and Big Data Engineering
Subtitle of host publicationImage and Video Data Engineering - 5th International Conference, IScIDE 2015, Revised Selected Papers
EditorsXiaofei He, Zhi-Hua Zhou, Xinbo Gao, Zhi-Yong Liu, Yanning Zhang, Baochuan Fu, Fuyuan Hu, Zhancheng Zhang
PublisherSpringer Verlag
Pages249-259
Number of pages11
ISBN (Print)9783319239873
DOIs
StatePublished - 2015
Event5th International Conference on Intelligence Science and Big Data Engineering, IScIDE 2015 - Suzhou, China
Duration: 14 Jun 201516 Jun 2015

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9242
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference5th International Conference on Intelligence Science and Big Data Engineering, IScIDE 2015
Country/TerritoryChina
CitySuzhou
Period14/06/1516/06/15

Keywords

  • Abnormal event
  • Crowd analysis
  • Double sparse representations
  • References

Fingerprint

Dive into the research topics of 'Abnormal crowd motion detection using double sparse representations'. Together they form a unique fingerprint.

Cite this