摘要
本文提出了一种基于FAST算法的无人机图像密集匹配,使用FAST算法提取两图像的特征点,并使用归一化互相关算法实现两特征点点集的匹配,然后根据初匹配的结果采用RANSAC算法鲁棒地估算出基本矩阵和单应矩阵;采用极线约束和单应约束通过归一化互相关关系稳健匹配出所有点。实验表明,该密集匹配方法结果令人满意。
A method of UAV image-dense matching based on FAST algorithm was proposed in this paper. Firstly, it is to extract image feature points using the FAST algorithm, and use the normalized cross-correlation relationship algorithm to match two feature point sets, then the fundamental and homographic matrixes were estimated by the robustness of the RANSAC algorithm according to the early matching results. All the points could be matched stably by the Epipolar constraint and Homography constraint through normalized cross-correlation relationship. The experiments showed that the matched results with the method were satisfactory.
出处
《测绘科学》
CSCD
北大核心
2013年第5期126-128,132,共4页
Science of Surveying and Mapping
基金
民用航天预研项目(Y1K00200KJ)
无人机作业控制系统软件研发(08Y02910KB)
欧盟第七框架项目(EU-FP7)(247608)
关键词
FAST算法
密集匹配
单应约束
极线约束
FAST algorithm
dense matching
epipolar constraint
homograpby constraint
