高分辨率遥感图像配准控制点均匀化算法

发布时间：2018-05-30 06:17

本文选题：高分辨率遥感影像配准 + 降采样预匹配　；参考：《遥感信息》2016年03期

【摘要】：针对多源遥感图像普遍具有数据量大、辐射差异大等特征,而现有的图像配准算法无法直接应用于遥感图像自动配准处理中的问题,综合考虑控制点的密度和分布,提出了一种高分辨率遥感图像自动配准算法。首先,将待配准图像和参考图像降采样到单机可以处理的大小,利用尺度不变特征变换(SIFT)算法建立降采样图像间的初始匹配;其次,将原始待配准图像按照网格分割为子图像,并利用初始匹配找到每幅子图像在参考图像上的对应子图像;再次,利用SIFT和极大稳定区域(MSER)特征点的空间互补性,在每一对子图像上提取大量特征点;最后,利用随机采样一致性(RANSAC)算法剔除误配后,采用基于最大团问题的贪心法进行控制点均匀化处理,进一步剔除冗余的控制点。与现有的基于SIFT特征和基于灰度的遥感图像配准算法相比,本算法在配准精度和控制点的分布均匀度等方面具有优越性。
[Abstract]:Multi-source remote sensing images generally have the characteristics of large amount of data and large difference of radiation, but the existing image registration algorithms can not be directly applied to the problems of automatic registration of remote sensing images. The density and distribution of control points are considered synthetically. An automatic registration algorithm for high resolution remote sensing images is proposed. First of all, the original image and reference image are de-sampled to the size that can be processed by a single machine, and the initial matching between the subsampled images is established by using the scale-invariant feature transform (sift) algorithm. Secondly, the original image is segmented into sub-images according to the grid. Using the initial matching to find the corresponding sub-image of each sub-image on the reference image; thirdly, using the spatial complementarity of the SIFT and the maximum stable region) feature points, extracting a large number of feature points on each pair of sub-images; finally, The random sampling consistency algorithm is used to eliminate the mismatch, then the greedy method based on the maximum cluster problem is used to homogenize the control points, and the redundant control points are further eliminated. Compared with the existing remote sensing image registration algorithms based on SIFT features and gray level, this algorithm has advantages in registration accuracy and distribution uniformity of control points.
【作者单位】：青海省第二测绘院;中国科学院遥感与数字地球研究所;
【基金】：国家自然科学基金(41301499)
【分类号】：TP751

【参考文献】