中文摘要：

针对航拍视频图像在光学变换和几何变换下特征匹配容易产生误匹配,以及匹配精度不一致等问题,提出一种结合不变特征和映射抑制的航拍视频配准算法。首先,利用Harris检测器在尺度变化下的局部区域内选择出最稳定角点,并利用分区检测方法提取出分布均匀的角点作为待配准点。然后,计算以特征点为中心邻域窗口的描述符,通过比较两幅待配准图像特征点邻域描述符的欧式距离,提取出初始匹配点对。对匹配对点集建立D-三角网图,利用映射抑制方法提取出精确度最高的少数匹配点,并对模型参数进行估计,完成视频帧间的配准。大量实验表明,所提方法对航拍视频图像配准具有较强的适应性且配准精度高。

英文摘要：

The Harris detector can produce false corners under common photometric transformations and geometric transfor- mations, and the matching points have different accuracy. This paper presents an aerial video image registration scheme by combining the invariant features and mapping restraint. Firstly, the locally most stable points are selected by the multi-reso- lution Harris detector, and a subarea detection method is presented to determine the well-distributed corners. The descrip- tors are established at each point in a circular neighborhood. Then Euclidean distance of the corner descriptors are compared to extract the initial feature point pair. Then the Delaunay triangulation of matching point pairs are computed, and the most ＂useful＂ matching points that best satisfy the cross-ratio invariants are presented to estimate the geometry transformation, and the image registration is completed. Experimental results demonstrate that the proposed algorithm can effectively offer a robust and accurate image registration for aerial video images.