基于亚像素红外与可见光图像配准方法  

Method of infrared and visible image registration based on sub-pixel detector

作　　者：朱智超[1] 闫钧华[1] 徐诚[1] 

机构地区：[1]南京航空航天大学航天学院,南京210016

出　　处：《国外电子测量技术》2011年第9期25-29,共5页Foreign Electronic Measurement Technology

基　　金：南京航空航天大学基本科研业务费专项科研项目(NS2010214)资助项目

摘　　要：为了提高不同分辨率的红外与可见光图像配准的精度,提出了一种基于亚像素纠正的红外与可见光自动配准方法。采用了Shi-Tomasi角点检测器和Taylor多项式将角点的坐标纠正到亚像素级,在六参数仿射变换模型基础上采用改进的Hausdorff距离进行精确匹配。与传统方法比较,该方法提高精度10%左右。实验表明,该方法配准精度高,鲁棒性好。In order to increase the accuracy of registration for different resolution of infrared and visible image,a method of infrared and visible image automatic registration is proposed based on sub-pixel correction.Shi-Tomasi corner detector and Taylor polynomial corrected accuracy of corners＇ coordinate are used to sub-pixel level,then on the base of 6 parameters affine transformation model uses modified Hausdorff distance for exact match.Compared with traditional method,this one increases accuracy about 10%.Experiment proves that the method is of high accuracy and robustness.

关 键 词：图像配准 亚像素 仿射变换 Hausdorf距离 红外图像 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]