任意指定地面复杂场景图像目标的去均值相关跟踪算法  

A Mean Removal Correlation Tracking Algorithm in the Arbitrary Specific Image Target of Complicated Ground Background

作　　者：江和平[1] 王平[1] 沈振康[1] 

机构地区：[1]国防科技大学ATR-2实验室,湖南长沙410073

出　　处：《信号处理》2006年第1期24-27,共4页Journal of Signal Processing

摘　　要：本文通过对相关系数作为相似性度量准则的匹配跟踪算法的改进,提出了去均值相关跟踪算法,采用网格搜索法进行粗匹配,以邻域法进行精匹配,减少了算法的计算量,满足机载成像光电吊舱系统目标跟踪的实时性要求。提出了自适应模板更新方法和修正的自适应模板更新方法,并对这两种方法进行仿真比较,实验表明,修正的自适应模板更新方法具有更高的跟踪精度和更好的跟踪稳定性。In the paper, a tradition correlation matching algorithm based on con＇elation coefficient has been improved and the mean removal correlation tracking algorithm was defined. With the use of the grid search method for the coarse matching and the use of the adjacent search method for the extractive matching, the algorithm improved greatly decrease operations and can be applied to airborne photoelectric real -time image tracking system. The adaptive template update method and the modificatory adaptive template update method is mentioned in the paper. The experimental results shows that the modificatory adaptive template change method is more tracking stability and tracking precision than the adaptive template change method.

关 键 词：去均值相关 目标跟踪 图像匹配 模板更新 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] O156.4[自动化与计算机技术—计算机科学与技术]