CBERS-02B星HR数据几何纠正模型研究

Study on Geometric Rectification Model of CBERS-02B HR Data

作为目前我国发射的分辨率最高的民用卫星,CBERS-02B星HR数据在大比例尺资源环境调查和基础地理信息更新等方面发挥着重大的作用。为了推广其应用,本文在不能获取卫星轨道星历参数和传感器参数的情况下,对覆盖两种不同地形类别的CBERS-02B星HR影像1级数据进行了多项式、投影变换、RPC等几种不同几何模型的纠正试验,并进行了精度分析与评定。本次试验结果表明:当控制点数量不多时,3阶多项式纠正精度相对较高,对于地形起伏较小的地区,所需的控制点数量为20个左右就可获取较高的精度,而对地形起伏较大的地区,控制点的数量则需要30个以上才能获得较高的精度;无足够控制点时RPC模型纠正精度不理想。

As the highest-resolution commercial satellite launched by China at present,CBERS-02B HR data plays an important role in the aspects including large-scale resource and environment investigation,updating of fundamental geographic information,etc.In order to promote the application of CBERS-02B data,in the absence of orbit ephemeris and camera parameters,CBERS-02B HR level 1 data covering two different kinds of terrain are rectified by using polynomial,projective transformation,RPC model,etc.and then a comparative analysis of the accuracy of these geometric rectification models were conducted.The experiments show that,when there is not many ground control points（GCPs）,the rectification accuracy of cubic polynomial is best among these models.And at the same time,20 GCPs are enough to achieve a high accuracy in lower topographic relief areas,while it needs more than 30 GCPs to get a high accuracy in higher topographic relief areas;the rectification accuracy of RPC model is unsatisfactory if without enough GCPs.