基于交通视频序列的多运动目标跟踪算法

来源期刊:中南大学学报(自然科学版)2010年第3期

论文作者:高韬 刘正光 张军

文章页码:1028 - 1036

关键词:多运动目标跟踪;运动识别;智能交通系统;车辆跟踪

Key words:multiple moving targets tracking; motion detection; intelligent transportation system; vehicle tracking

摘    要:针对智能交通领域对自适应多运动目标跟踪的广泛需求,提出一种新型的基于交通视频序列的多目标跟踪算法。通过Marr小波概率核函数生成静态背景,并结合当前帧在B/RDWT(Binary/redundant discrete wavelet transforms)域进行多运动目标识别,同时采用边缘阴影剔除算法去除阴影的干扰。运动跟踪采用SI_P(SIFT-particle)粒子滤波算法,并结合改进的均值漂移(mean-shift)法获得运动目标的准确跟踪窗口。采用队列链表法记录多运动目标之间的数据关联,在提高识别准确率的同时降低运算的复杂度。算法采用VC++6.0实现,通过实际道路测试,研究结果表明:SI_P粒子滤波算法与传统算法相比,平均时耗只多0.15 s,跟踪窗口尺度可自适应变化,并且该算法对于多运动目标识别跟踪具有更优越的实时性和抗遮挡性。

Abstract: For the demanding of adaptive multiple moving targets tracking in intelligent transportation field, a new type of traffic video widely based multi-target tracking algorithm was presented. Background was modeled by Marr wavelet probability kernel function and a background subtraction technique based on binary/redundant discrete wavelet transforms was introduced to detect multiple moving targets. After obtaining the foreground, shadow was eliminated by an edge detection method. A type of SI_P (SIFT-particle) filter combined with improved mean-shift method was used for video tracking, and tracking window adaptively changed its scale according to the size of target. A Queue chain method was used to record data association among different targets, which could improve the detection accuracy and reduce the complexity. The software is VC++6.0, and by actual road tests, the average runtime of SI_P algorithm is only 0.15 s more than that of traditional algorithm; the scale of tracking window can adaptively changes. The algorithm tracks multi-target with a better performance of real time and mutual occlusion robustness; it can be used in intelligent traffic monitoring with extensive application prospect.

基金信息:国家自然科学基金资助项目
天津市自然科学基金资助项目
天津市科技支撑计划重点项目基金资助项目
天津市公安交通局科研基金资助项目

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号