一种基于Snake模型的多目标跟踪算法  

A Multiple Objects Tracking Algorithm Based on Snake Model

作　　者：刘皓挺[1] 姜国华[1] 王丽[1] 

机构地区：[1]航天医学工程研究所,北京100094

出　　处：《计算机工程与应用》2006年第7期76-79,共4页Computer Engineering and Applications

摘　　要：在视频图像处理中,由于Snake模型运算量大且易受背景物体或噪声的干扰,因此很少被用来完成目标跟踪任务。论文主要针对高空摄像机追踪地面多个运动目标的具体情况,通过研究Snake模型各个能量函数的特点,提出了一种结合拆分模板匹配技术、Kalman滤波技术及Snake模型的多目标跟踪算法。与同类算法相比,该算法具有建模简单、分割精度较高的特点。Snake model is not applied widely because of its process of time consuming and characteristics of poorly resisting noise or background objects.In this paper an algorithm combining with split template and Kalman filter is introduced to improve the performance of Snake model.The approach can be used to track multiple objects in aerial image sequences or remotely sensed image ones.Experimental results show that if occlusions are avoided the approach can track multiple objects well on uncluttered background.

关 键 词：SNAKE 多目标跟踪 KALMAN滤波 拆分模板 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]