基于运动补偿的Snake视频对象跟踪算法  

Video object tracking method based on motion compensation Snake model

作　　者：郭礼华[1] 李建华[1] 杨树堂[1] 

机构地区：[1]上海交通大学电子信息与电气工程学院,上海200030

出　　处：《红外与激光工程》2005年第1期93-97,共5页Infrared and Laser Engineering

基　　金：国家863高科技资助项目(2002AA145090)

摘　　要：当对象移动幅度大时,Snake视频对象跟踪算法中的曲线迭代过程易陷入局部最小,为此,提出一种运动补偿预处理的解决方法。此方法预先估计对象的运动信息,然后对Snake曲线的初始轮廓位置进行运动补偿,最后进行Snake跟踪。实验表明,这种方法不仅跟踪效果好,而且能有效地减少Snake曲线演化的迭代次数。If a somewhat fast moving object exists in successive images, Snake's nodes may fall into the local minima in their motion to the new positions of the target object in next image. In this paper, Snake tracking algorithm based on the motion compensation is presented to overcome this problem. Firstly, the motion information of object is analyzed. Secondly, the object's position is made compensated according to this motion information.Finally, the object is tracked by Snake model. The final experimental results prove that this method performances a good tracking result and can greatly decrease the number of the Snake's contour evolvement.

关 键 词：SNAKE模型 主动轮廓模型 视频对象跟踪 运动补偿 运动估计 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]