WorldView-1影像RFM多项式平差模型及其精度分析

简要给出了WorldView-1卫星及传感器的有关性能参数,详细论述了可用于WorldView-1影像数据几何处理的多种RFM(Rational Function Model)多项式平差模型。然后,使用云南地区的WorldView-1影像数据进行了有关试验研究,分别采用九种零阶、一阶和二阶多项式平差模型,在像方改正RFM参数的系统误差。结果表明,当地面控制点在精度、数量与分布等方面质量较好时,各种模型精度基本一致,平面精度可达到1.6个像素(0.9m)左右。此外,本文重点研究了在控制点质量较低的情况下,各种RFM多项式平差模型的改正精度。发现随着控制质量的降低,二阶多项式和一阶多项式改正的精度有显著下降,但零阶多项式改正的精度基本稳定不变。因此,本文认为零阶多项式模型,即平移模型,是一种最简单、适应性最好、精度最高的用于处理WorldView-1影像数据的平差模型。

Accuracy Analysis of Polynomial RFM Adjustment Models for WorldView-1 Imagery

In this paper,the specifications of WorldView-1 and its payloads are introduced,and nine polynomial adjustment models（from zero order to second order） used to improve the vendor provided RFM（Rational Function Model） accuracy are discussed in detail.WorldView-1 images in Yunnan Province are used to test the correction accuracy of these adjustment models using 81 GPS points.Results show that the accuracy of nine modes is similar when using well distributed GCPs with high accuracy and the planar RMS errors can reach 1.6 pixels（0.9 meter）.The correction accuracy of polynomial RFM adjustment models using sparse,badly-distributed,or low accuracy GCPs is studied emphatically in this paper.Research results show that with the degradation of GCPs accuracy,the errors of Independent Check Points（ICPs）,using zero order polynomial adjustment model,remain stable,but that of one order and two order polynomial adjustment models degrade distinctly.In conclusion,the zero-order polynomial RFM adjustment is the simplest,most adaptive and has highest accuracy for WorldView-1 image,and it is recommended for the difficult area mapping with sparse or badly-distributed GCPs.