OBJECT TRACKING BY MULTI-DEGREES OF FREEDOM MEAN SHIFT PROCEDURE COMBINED WITH THE KALMAN PARTICLE FILTER ALGORITHM
The paper begins with an analysis of the shortcomings of existing methods. We aim to overcome these shortcomings with our improved mean shift algorithm in which we introduce two distinguishing features: the bandwidth matrix and the target angle. We first introduce the bandwidth matrix mean shift procedure. Then we describe the target by introducing the target rectangle, which provides two positions coordinates of the centre point, the horizontal axis, the vertical axis and the target angle, altogether five degrees of freedom.Target angle is used to accommodate the rotation of objects while the two axes determine the size in two independent directions. Furthermore, we incorporate the Kalman Particle Filter (KPF) into our tracking framework to cope with a temporal occlusion of the objects. Experiments with several real worlds sequences indicate our new methods capability to adapt to any combinations of the targets rotation, zooming and translation. With better description of the object it achieves much better precision.
Mean Shift Tracking of Objects in Image Sequences Target angle Bandwidth Matrix Adaptability Kalman Particle Filter
JING-PING JIA QING WANG YAN-MEI CHAI RONG-CHUN ZHAO
School of Computer Science and Engineering Northwestern Polytechnical University, Xian 710072, China
国际会议
2006 International Conference on Machine Learning and Cybernetics(IEEE第五届机器学习与控制论坛)
大连
英文
3793-3797
2006-08-13（万方平台首次上网日期，不代表论文的发表时间）