具有SIFT描述的Harris角点多源图像配准  

Registration of Multi-sensor Images Based on Harris Corner with SIFT Description

作　　者：芮挺[1] 张升奡[1] 周遊[2] 孙峥[1] 曹鹏[1] 

机构地区：[1]解放军理工大学工程兵工程学院,南京210007 [2]江苏经贸职业技术学院,南京210007

出　　处：《光电工程》2012年第8期26-31,共6页Opto-Electronic Engineering

基　　金：国家自然科学基金资助(61070108);解放军理工大学工程兵学院基金资助

摘　　要：多源传感器成像原理的差异给图像配准带来了很大困难,本文针对红外与可见光图像配准提出了一种具有SIFT描述特征的Harris角点多源图像配准算法。首先建立多尺度空间,以多尺度空间检测尺度不变的Harris角点作为特征点;然后通过改进SIFT对特征点的描述方法,采用圆环结构算子对Harris角点进行类SIFT的特征描述;最后利用双向最近邻方法进行匹配,通过最小二乘法实现图像的配准。实验证实了算法配准的精确性、快速性和稳定性,具有较好的配准效果。Multi-sensorimage registration is very difficult because of their different imaging principles. Referring to the registration of infrared and visible images, a registration algorithm is proposed based on the Harris comer with SIFT description. First, the scale spaces were built, and the scale-invariant Harris comers were selected as feature points. Then, by improving SIFT description method of feature points, ring operator was used to describe feature points. Finally, feature points were matched by the two-way nearest neighbor algorithm, and least squares method was employed to find optimal solutions to affine transform equations. Experimental results show that the registration algorithm is accurate, efficient and stable, and has a good registration result.

关 键 词：多源图像 多尺度Harris角点 SIFT描述 图像配准 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]