基于主成分分析的目标跟踪方法研究  

Research of Object Tracking Method Based on Principal Component Analysis

作　　者：邢志胜[1] 张桂林[1] 王冼[1] 

机构地区：[1]华中科技大学图像识别与人工智能研究所图像信息处理与智能控制教育部重点实验室,武汉430074

出　　处：《计算机与数字工程》2002年第5期14-17,共4页Computer & Digital Engineering

摘　　要：在成像跟踪系统中,实时图与预存的参考图之间往往存在着巨大的差异。传统的匹配定位方法不能很好 的克服这些差异给跟踪系统带来的影响,使得目标漂移,跟踪失败。本文运用主成分分析方法动态构造模板, 运用仿射变换的方法匹配定位,从而实现稳定的跟踪。试验表明跟踪算法具有较高的鲁棒性。In the image tracking system, there is difference between the real time images and the reference images which are stored in advance. Traditional matching methods can not favorably cope with the difference that affects the tracking system , which makes the objects shifting and the tracking failing. It is proposed to reconstruct model dynamically and using the method of affine transform to match. So we can realize stable tracking. The experimental results are presented to verify the robustness of the tracking method.

关 键 词：主成分分析 目标跟踪 特征子空间 图像匹配 仿射变换 图像处理 

分 类 号：TN911.73[电子电信—通信与信息系统] TP391.41[电子电信—信息与通信工程]