Moving people detection in dynamic scenes by stereo vision

Tao Zhuo, Yanning Zhang, Tao Yang, Xiaoqiang Zhang

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

Abstract

Robust people localization on a moving robot platform is an important and challenge research topic. In this paper, we present a novel moving people detection approach and system on a mobile robot platform. The proposed method mainly contains two parts: (1) A Histograms of Oriented Gradients (HOG) based detector is adopted to detect the human candidates in the dynamic scene; (2) Geometric constraints are computed to handle the ghost problem, including the depth information from a stereo camera and the external parameters of the camera calibration. To evaluate the proposed approach, a robust people detection system is developed on a moving robot platform. Extensive experiment results with challenge indoor and outdoor scenarios demonstrate the robustness and efficiency of our approach.

Original languageEnglish
Title of host publicationIntelligent Science and Intelligent Data Engineering - Second Sino-Foreign-Interchange Workshop, IScIDE 2011, Revised Selected Papers
Pages506-512
Number of pages7
DOIs
StatePublished - 2012
Event2nd Sino-Foreign-Interchange Workshop on Intelligent Science and Intelligent Data Engineering, IScIDE 2011 - Xi'an, China
Duration: 23 Oct 201125 Oct 2011

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7202 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference2nd Sino-Foreign-Interchange Workshop on Intelligent Science and Intelligent Data Engineering, IScIDE 2011
Country/TerritoryChina
CityXi'an
Period23/10/1125/10/11

Keywords

  • Histograms of Oriented Gradients
  • moving people detection
  • stereo vision

Fingerprint

Dive into the research topics of 'Moving people detection in dynamic scenes by stereo vision'. Together they form a unique fingerprint.

Cite this