摘要
针对大失配多传感器图像,提出了一种基于SIFT(scale invariant keypoints)和Harris-Affine(H-A)互补不变特征匹配的自动配准算法.算法应用SIFT和H-A两种具有互补特性的局部不变特征,根据最近邻特征点距离与次近邻特征点距离之比确定初始匹配点对,然后利用马氏距离的仿射不变性删除误匹配特征点对,据此求取2幅源图像间的仿射变换参数.使用估计的变换矩阵把待配准图像上的所有点映射到参考图像,并对其进行重采样,实现图像的配准.实验结果表明:该算法能够快速高精度实现大失配图像的自动配准.
An automatic image registration algorithm based on the complementary SIFT and HarrisAffine （H-A） local invariant features was proposed for large misalignment multi-sensor images. In this algorithm, SIFT features were complemented with H-A features and the ratio of the first and second nearest neighbor distance were used to setup the initial correspondences. The affine invariant of Mahalannobis distance was used to remove the mismatched feature points. With this correspondence of the points, the affine matrix between two different images could be determined. All points in the sensed image were mapped to the reference using the estimated transformation matrix and the corresponding gray level was assigned by re-sampling the image in the sensed image. Experiments demonstrated the feasibility of this method.
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2008年第8期13-16,共4页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家高技术研究发展计划资助项目(2006AA01Z129)
国家重点基础研究发展计划资助项目(2007CB311005)
河南省教育厅自然科学基础研究计划资助项目(2007510023)
985工程二期科技创新平台项目
福建省自然科学基金资助项目(A0710020)
关键词
图像配准
SIFT特征
Harris-Affine特征
马氏距离
image registration
scale invarians keypoints （SIFT） feature
Harris-Affine （H-A） feature
Mahalannobis distance
