摘要
目标跟踪算法中往往需要有效的特征对目标进行跟踪识别。本文提出了随机边缘特征的概念,对其物理意义及匹配算法进行了阐述,并将其应用到运动车辆跟踪算法中。实验结果表明,该算法能够对序列图像中的多个运动车辆进行跟踪识别,对车辆的尺度、视角、光线及运动状态的变化等不利因素有较好的适应性。
An effective feature is often needed for target tracking algorithms to track and recognize targets. This paper proposes the concept of random edge feature.The physical significance and matching method of this feature are expounded. This feature has been applied in vehicle tracking algorithms. Experimental results show that the proposed algorithm can track and recognize multiple moving vehicles in the image sequence, and it can adapt well to the disadvantageous factors such as the changing of scales, angles of view point,rays, and moving states of vehicles.
出处
《计算机工程与科学》
CSCD
2005年第3期35-37,40,共4页
Computer Engineering & Science
