Georgian Electronic Scientific Journal: Computer Science and Telecommunications 2009|No.1(18)
Application research of Mean Shift Algorithm in the embedded for
real-time color Target Tracking
Shuang Qiao1, Xiufen Li2 , Yan Zhu3
1 School of physics Northeast Normal University, Changchun, 130024, firstname.lastname@example.org
2 School of physics Northeast Normal University, Changchun, 130024
3 School of physics Northeast Normal University, Changchun, 130024
In order to meet the real-time performance requirements of Mean Shift Algorithm,
this paper proposes a new kernel-based quantitative method and integralization method
on calculating the weights. This method makes the complex floating-point operation to
be a integer computation, which can reduce the computation. We Design a closed-loop
tracking network which composted of an embedded target tracking platform which bases
on a high-performance TMS320C6418 DSP chip with large-scale gate array, and a
servo system with two degrees of freedom. The miss distance of the object (the distance
the goal from the center of the view field)is the feedback signal to control the servo
system to move the camera. So that ,the goal will be near the center of view field.
Experiment result shows that the system has a good stability.
Keywords: closed-loop tracking color; characteristics; mean shift algorithm
How to effectively track the useful goal in the sequence video, is an extremely challenge topic
in the computer vision, and it has the widespread application in the monitoring, video coding and
the war industry. In the initiative visual system, we not only need to get the trajectory of target fast
and accurately, but also need to control the tracking turntable in the tracking process to maintain the
target near the center field of view to realize a continuous tracking .
Color feature is applied widely in the field of computer vision,