一种新的视频流动态目标跟踪算法.pdfVIP

优先出版 计 算 机 应 用 研 究 第32 卷 一种新的视频流动态目标跟踪算法 宗 薇 (外交学院 信息中心，北京，100037) 摘 要：为了提高视频图像序列动态目标跟踪过程的跟踪效率，在分析现有算法的基础上，提出了一种基于卡尔曼预测 和协方差模块更新相结合的动态目标跟踪新算法。通过卡尔曼预测，可以得到图像序列中下一帧的动态有效区域，实现 了对目标区域的实时跟踪；同时，目标协方差矩阵的更新用于目标匹配区域的预测，提高了整个跟踪算法的抗干扰性能 和鲁棒性。实验结果表明，与现有的基于静态模板的协方差跟踪算法相比，该算法在实时跟踪性能方面具有更好的跟踪 效率。该方法用于视频图像序列动态目标跟踪是可行的、有效的。 关键词：目标跟踪；视频流；协方差；卡尔曼预测 中图分类号：TP391 文献标志码：A New video stream dynamic target tracking algorithm ZONG Wei (Computer Center of China Foreign Affairs University Information Center, Beijing 100037, China) Abstract: In order to improve tracking efficiency of video sequences dynamic target tracking process, proposing a new dynamic target tracking algorithm based on combination of Kalman prediction and covariance module updates after analysis of the existing algorithms. According Kalman prediction, dynamic effective area of the next frame in the image sequence is obtained. Meanwhile, target covariance matrix updates is applied to forecasts of target matching area. Anti-jamming performance and robustness of the whole tracking algorithm are improved. Experimental results show that, compared with the existing covariance tracking algorithm based on static template, it has better tracking efficiency in real-time tracking performance. The method used for video image sequence moving target tracking is feasible and effective. Key Words: target tracking; video flowing; covariance; Kalman prediction 量的双不变性质引入到其中，得出一种黎曼指数映射和黎曼均 0 引言