基于SIFT特征目标跟踪算法研究  

Research on Object Tracking Algorithm Based on SIFT

作　　者：蔺海峰[1] 马宇峰[1] 宋涛[1] 

机构地区：[1]西安通信学院,西安710106

出　　处：《自动化学报》2010年第8期1204-1208,共5页Acta Automatica Sinica

摘　　要：在尺度不变特征变换(Scale invariant feature transform,SIFT)特征匹配算法的基础上,提出了一种基于累积特征的多目标的跟踪算法,通过对目标的SIFT特征进行实时更新来去除由噪声(或形变)带来的"过时"特征信息,保证了特征的稳定,提高了匹配准确度.实验结果表明,本算法能够有效处理目标由于旋转、形变而导致跟踪性能下降甚至跟踪目标丢失的问题,同时对跟踪过程中多目标的关联,也具有较好的鲁棒性.Based on the scale invariant feature transform （SIFT）,a novel motion-tracking algorithm for multi-targets utilizing the feature reserving priority of preference is proposed.The SIFT features of an object are updated in real time to store the stable features of a recent frame.Thus,it can realize the stable tracking of multi-objects by feature reserving priority of preference instead of prior information.Experimental results show that this method can not only handle the problems of target losing effciently,which are induced by object s rotation and translation,but also has nice robustness to the conjunction of multi-targets in the process of object tracking.

关 键 词：多目标跟踪 尺度不变特征变换 非刚性形变 稳定性 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]