基于特征匹配的地图图像自动配准技术研究

本文针对地图中的特征点，提出了一种基于广义特征点的图像自动配准方法，将特征点从单纯的点拓展到特征区域。以Moravec算子结合其他特征约束条件来自动搜索广义特征点。分别对两幅图像提取广义特征点后，利用基于根均方误差和交叉相关的两级匹配算法完成同名控制点的建立。并以局部加权直线拟合方法来校正图像的几何畸变。最后建立两幅图像之间的函数映射关系，完成图像的配准。实验结果证明了该方法的有效性。该方法可用于校正近景面地图影像的几何畸变和遥感图像的局部几何畸变。

A Feature-based Technique and Algorithm to Automated Image-to-image Merge

In this paper,a feature-extended manner is presented to automated image-to-image merge.Firstly the feature point extended from a simple one to a feature region.The kind of broad feature points can be detected automatically through Moravec operator and other feature restrictions.When feature point sets are extracted respectively from the two images,their correspondence between two point sets is established by a two-stage matching algorithm,RMSE(Root Mean Square Error)and cross correlation.Then the method of local weighted mean is used to register the image distortions.Finally,the function mapping relation between the two images is established to form image-to-image automated merge.Experiments showed that this method can successively register the local geometric distortions not only for the map images but also for the remote sense images.