采用非线性尺度空间滤波和SIFT的遥感影像配准方法  

Remotely Sensed Imagery Registration Based on Nonlinear Scale-Space Filtering and SIFT

作　　者：施文灶[1,2,3,4] 毛政元[1,3,4] 

机构地区：[1]福州大学空间数据挖掘与信息共享教育部重点实验室,福建福州350002 [2]福建师范大学光电与信息工程学院,福建福州350108 [3]福州大学地理空间信息技术国家地方联合工程研究中心,福建福州350002 [4]福州大学福建省空间信息工程研究中心,福建福州350002

出　　处：《华侨大学学报（自然科学版）》2016年第1期38-42,共5页Journal of Huaqiao University(Natural Science)

基　　金：国家自然科学基金资助项目(61275006;41201427);"十二五"国家科技支撑计划项目(2013BAC08B02-01);国家重点基础研究发展计划项目(2006CB708306);福建省教育厅科研基金资助项目(JB14038)

摘　　要：针对传统点特征匹配算法存在运算时间长和配准精度低的问题,提出一种基于非线性尺度空间滤波和尺度不变特征转换(SIFT)点特征配准算法.首先,通过非线性尺度空间滤波对基准影像和待配准影像分别进行预处理,保留其边缘信息并去除噪声.其次,采用SIFT算法对预处理后的两幅影像进行特征点提取,通过最近邻和次近邻的欧式距离比值法进行双向匹配,得到匹配特征点.最后,对待配准影像进行仿射变换.结果表明:该方法的总体运行时间比传统SIFT点特征配准算法降低63.2%,且配准精度大幅提高.To solve the problems of long executing time and low registration accuracy of the traditional point feature matching algorithm, this article proposed an improved scale-invariant feature transform （SIFT） point feature matching ap proach based on the nonlinear scale-space filtering. Firstly, the reference image and the to-be-registered one were respec- tively preprocessed with the nonlinear scale-space filter filtering. Secondly, feature points were extracted from the two images by means of the SIFT algorithm. Then, matched feature points were obtained through a bilateral matching by the ratio of Euclidean distances of the nearest neighbor to that of the next nearest one. Finally, an affine transformation was carried out to the to-be-registered image. Experimental results show that the executing time of the proposed method re duces 63.2% compared with the traditional SIFT point feature matching algorithm, and the registration accuracy is signif- icantly improved.

关 键 词：遥感影像 非线性尺度空间滤波 尺度不变特征转换 配准 仿射变换 

分 类 号：P237[天文地球—摄影测量与遥感]