Digital

Vehicle detection methods for surveillance applications

Publikation aus Digital

Sidla O., Wildling E., Lipetski Y.

SPIE Intelligent Robots and Computer Vision XXIV: Algorithms, Techniques, and Active Vision; Boston, USA Oct 2006 , 2006

Abstract:

The efficient monitoring of traffic flow as well as related surveillance
 and detection applications demand an increasingly robust recognition
 of vehicles in image and video data. This paper describes two different
 methods for vehicle detection in real world situations: Principal
 Component Analysis and the Histogram of Gradients principle. Both
 methods are described and their detection capabilities as well as
 advantages and disadvantages are compared. A large sample dataset
 which contains images of cars from the backside and frontside in
 day and night conditions is the basis for creating and optimizing
 both variants of the proposed algorithms. The resulting two detectors
 allow recognition of vehicles in frontal view +- 30 deg and views
 from behind +- 30 deg. The paper demonstrates that both detection
 methods can operate effectively even under difficult lighting situations
 with high detection rates and a low number of false positives.