Title :
Kernel based multiple targets tracking
Author :
Wang, Yongzhong ; Pan, Quan ; Liang, Yan
Author_Institution :
Sch. of Autom., Hangzhou Dianzi Univ., Hangzhou, China
Abstract :
In multiple targets tracking, the split, mergence or occlusion of objects will deteriorate the tracking performance, so a multiple targets tracking method based on kernel is proposed. Making use of the defined matching measurement rule, a matching matrix is constructed according to the detection results based on a Gaussian mixture model and the kernel-based tracking results, and then the spilt and merging operation is completed to realize tracking target initialization, deletion, split, merging and occlusion. The experimental results show that the proposed method can reliably track multiple targets under occlusion in real time.
Keywords :
Gaussian processes; image matching; matrix algebra; target tracking; Gaussian mixture model; kernel based multiple targets tracking; matching matrix; matching measurement rule; spilt and merging operation; Automation; Educational institutions; Intelligent control; Kernel; Merging; Target tracking; Visualization; kernel-based tracking; merge; multiple targets; split;
Conference_Titel :
Intelligent Control and Automation (WCICA), 2010 8th World Congress on
Conference_Location :
Jinan
Print_ISBN :
978-1-4244-6712-9
DOI :
10.1109/WCICA.2010.5554660