Title:	OCCLUSION HANDLING FOR PEDESTRIAN TRACKING USING PARTIAL OBJECT TEMPLATE-BASED COMPONENT PARTICLE FILTER
Author(s):	Daw-Tung Lin, Yen-Hsiang Chang
ISBN:	978-972-8939-89-2
Editors:	Yingcai Xiao
Year:	2013
Edition:	Single
Keywords:	Pedestrian tracking, occlusion handling, video surveillance, template-based component matching, Particle Filter.
Type:	Full Paper
First Page:	19
Last Page:	26
Language:	English
Cover:
Full Contents:	click to dowload
Paper Abstract:	Pedestrian tracking plays an important role in the realm of security and intelligent video surveillance. Occlusion handling is a challenging issue in tracking multi-people. Therefore, adaptive and advanced solutions are required to fulfill the accurate tracking task and to analyze the pedestrian behavior for specific video surveillance purpose. This paper presents a novel method and address the problem of tracking and evaluating the number of people in complex scenes with occlusion conditions. In this work, collaboration of component-based human shape template and Particle Filter is developed and the ability of handling object occlusion is improved. The proposed system is capable of tracking specific person in real-time and handle multiple objects occlusion. The occlusion situation is predicted by using Kalman Filter and then each object is continuously tracked by the component shape template Particle Filter. Experimental results show that our algorithm is feasible and stable. The proposed tracking algorithm achieves an accuracy of up to 99.7%. The proposed approach outperforms that of the other methods for all test video datasets. Our low false negative rate reveals that the proposed tracking method is robust and superior in occlusion handling.
	Go Back