一种带几何约束的大幅面遥感影像自动快速配准方法

A Rapid Image Registration Method Based on Restricted GeometryConstraints for Large-Size Remote Sensing Image

摘要: 针对大幅面遥感影像的配准需求,提出了一种带有几何约束的大幅面遥感影像快速自动配准方法。该方法首先结合遥感影像空间地理信息构建影像粗配准模型;然后在粗配准模型的约束下,通过对大幅面遥感影像进行规则格网划分,结合改进的SIFT算法和并行策略实现影像精配准。实验结果表明,本文方法能提高影像的配准效率,且配准精度较高,适用于大幅面遥感影像之间的配准。

Abstract: Considering the registration of large-size remote sensing images,a rapid image registrationmethod based on restricted geometry constraints is proposed.Firstly,the coarse image registrationcombining geography information is introduced for remote sensing image.Then,by dividing the large-size image into grid images,fine registration is done combining an improved SIFT algorithm and paral-lel architecture under the constraints of a coarse registration model.Experimental results show thatthe proposed method not only can improve the efficiency of image registration,but also can processhigher registration accurately,and applicable for registering large-size remote sensing images.