Total views : 434

A Geometric Approach to Video Surveillance

Affiliations

  • Department of CSE, Kongju National University, Korea, Republic of

Abstract


Intrusion detection is one of the important functions for smart video surveillance. Presented in this paper is a geometric approach to intrusion detection. Our approach consists of a geometric flow and an image flow. Given a 3D global boundary for a wide area requiring multi-CCTVs, it is transformed into the 2D local boundary for each camera in the geometric flow. The bounding box for a moving one is computed in the image flow. Finally, whether one intrudes a guard zone can be decided by calculating their intersection. By experiments, we could find our approach is robust and adaptive in multi- CCTVs environment where cameras are replaceable.

Keywords

Bounding Box, Geometric Approach, Global Boundary, Intrusion Detection, Local Boundary

Full Text:

 |  (PDF views: 203)

References


  • Free JA. Automatic video surveillance with intelligent scene monitoringand intruder detection. Security Technology. 30th Annual 1996 International Carnahan Conference; 1996 Oct 2-4; Lexington,Kentucky, USA. p. 89–94.
  • Oh S. Train detection in railway platform area using image processingtechnology. J of the Korea Academic-Industrial CooperationSociety. 2012;13(12):6098–104.
  • Won J. A study on the adaptive object recognition in safe zonemonitoring. The Korean Society for Railway. 2010; 7(1):124–8.
  • Available from: http://www.axis.com/solutions
  • Available from: http://www.i2vsys.com/products
  • Kim J, Maeng S. A map-based boundary method for video surveillance. J of the Korea Academic-Industrial Cooperation. 2014 Jan; 15(1):418–24.
  • Mount DM. Computational geometry [Lecture Note]. Department ofComputer Science: University of Maryland;2007.
  • Kim D. Open CV computer vision programming. Gamae; 2013.
  • Joo W. 3D computer graphics learning with Open GL. S Korea: Hanbit Academic Inc; 2013.

Refbacks

  • There are currently no refbacks.


Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 License.