摘要
提出了一种基于跟踪微分器(TD)的目标跟踪算法,该算法由跟踪微分器和序贯相似性匹配算法(SSDA)组成。假设目标在视频图像序列第一帧中的位置已知,使用跟踪微分器,可以预测出目标在下一帧图像中的位置,然后再使用SSDA算法,在预测位置的邻域内进行搜索,从而获得目标的准确位置。仿真实验证明基于跟踪微分器的模板匹配算法减少了搜索位置,提高了算法的效率,同时也克服了传统的卡尔曼滤波需要目标运动模型已知的缺点。
Target tracking algorithm based on Tracking Differentiator （TD） is presented. This algorithm consists of TD and SSDA algorithms. Assuming target location in video image of first frame is known, by use of TD, target location in video image of next frame can be predicted. Then by using SSDA algorithm, searching over neighborhoods of prediction position, correct location of target can be obtained. Simulation test showed that module compatible algorithm based on TD can decrease searching position, improve effectiveness of algorithm, and overcome drawback such as that traditional kalman filtration need movement model of target.
出处
《火炮发射与控制学报》
北大核心
2006年第3期6-9,14,共5页
Journal of Gun Launch & Control
基金
国家自然科学基金资助(60175010)。
