遥感图像与GIS数据的配准是遥感应用、遥感与GIS集成的基础和关键.目前的方式大多是通过人工选取一点数量的控制点,通过几何纠正模型把遥感图像纳入到GIS坐标系统中,同时改正遥感图像的几何畸变.随着遥感技术的发展,传统人工作业方法无论是从效率还是从质量上都无法满足日益增长的应用需求,因此需要研究实用化的遥感影像与GIS数据的自动配准方法.本文从实用化的角度出发,在分析特征匹配理论的基础上,提出了一种基于面状地物多边形特征的遥感影像与GIS数据自动配准方法.该方法不同于一般的特征匹配方法之处在于特征提取与特征匹配同步解求.本文对新方法的理论进行了详细论述,最后对上海浦东地区的QUICKBIRD影像和1:2000 GIS数据进行了配准实验,对比传统方法计算结果,新方法具有明显优势.
Matching remote sensing imagery and GIS data is very important and a key step for applications of the imagery and integration of the imagery and GIS. The general approach includes following steps: control points selection, computation of rectification model parameters of the image and rectification of the whole image pixel by pixel. But the processing is very time-consuming when performed manually. To improve the processing efficiency, we present a new method based on area features. Our approach, guided by GIS knowledge, iteratively extracts an area feature and matches area features simultaneously. The matching experiment, which uses a QUBICKBIRD image and 1 : 2000 GIS data in Pudong, Shanghai is presented. Comparison of the results of our approach and that of the traditional method indicates that the new approach can obtain better accuracy with limited manual operations.