ETH Zuerich - Startseite
Professur für CAAD

 


caadlogo.gif

Video Tracking


For various services in a building it is advantageous to monitor a persons whereabouts and sometimes even a track of the persons movement. For a Ph D thesis CAAD collaborates on use of a video tracker with the Computer Vision Lab (http://www.vision.ee.ethz.ch/).

The tracking algorithm by CVL uses a background segmentation as base. A picture showing the empty scene is reduced from any consecutively arriving picture. The resulting difference is an object or person that has not been there before. The geometric information on the detected and tracked object is available as bounding box for other applications. Torsten Spindler of CAAD checks the trackers behaviour in different situations. What happens with a semi variable background, e.g. a window with birds flying around or trees rocking in the wind. What happens with mirror images on semi-mirroring surfaces as windows. The results of these expirements are given to CVL to further improve the tracker. Currently tracking is limited to one object and CVLs plans are to expand this to multiple objects even when located in crowds of people.

Tracker.jpg

Revision r1.1 - 14 Sep 2005 - 08:03 - TorstenSpindler
Parents: ArticleFrontPage
Copyright © 1999-2003 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.