Adaptive homochromous disturbance elimination and feature selection based mean-shift vehicle tracking method

Jie Ding*, Bo Lei, Pu Hong, Chensheng Wang

*Corresponding author for this work

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

1 Citation (Scopus)

Abstract

This paper introduces a novel method to adaptively diminish the effects of disturbance in the airborne camera shooting traffic video. Based on the moving vector of the tracked vehicle, a search area in the next frame is predicted, which is the area of interest (AOI) to the mean-shift method. Background color estimation is performed according to the previous tracking, which is used to judge whether there is possible disturbance in the predicted search area in the next frame. Without disturbance, the difference image of vehicle and background could be used as input features to the mean-shift algorithm; with disturbance, the histogram of colors in the predict area is calculated to find the most and second disturbing color. Experiments proved this method could diminish or eliminate the effects of homochromous disturbance and lead to more precise and more robust tracking.

Original languageEnglish
Title of host publication2011 International Conference on Optical Instruments and Technology: Optoelectronic Imaging and Processing Technology
Pages1-12
Number of pages12
Volume8200
DOIs
Publication statusPublished - 2011
Event2011 International Conference on Optical Instruments and Technology: Optoelectronic Imaging and Processing Technology - Beijing, China
Duration: 6 Nov 20119 Nov 2011

Other

Other2011 International Conference on Optical Instruments and Technology: Optoelectronic Imaging and Processing Technology
CountryChina
CityBeijing
Period6/11/119/11/11

Keywords

  • Adaptive feature selection
  • Airborne video
  • Homochromous disturbance
  • Mean-shift
  • Vehicle tracking

Fingerprint Dive into the research topics of 'Adaptive homochromous disturbance elimination and feature selection based mean-shift vehicle tracking method'. Together they form a unique fingerprint.

Cite this