Video-based Vehicle Tracking Considering Occlusion

K. Uchimura and K. Matsushima (Japan)


Vehicle tracking, Occlusion, Template matching, Intelligent Transport Systems


This paper addresses the new methods of vehicle track ing and traffic flow analyzing. Partial or full occlusion is the main factor that affects the performance of tracking al gorithm. It becomes more critical in the ITS (Intelligent Transport Systems) applications like traffic flow measure ment and visual guidance in autonomous driving. In this paper, we present a new tracking algorithm based on the concept of tracking cycle and deformable template match ing. Vehicle detection is performed by the adaptive back ground differential processing. Additional traffic flow mea suring technique is also presented to calculate the so-called Traffic Parameters like traffic volume, speed and space oc cupancy. The algorithms are validated with the complex traffic stream video taken by a surveillance CCD camera fixed on the pedestrian bridge.

Important Links:

Go Back