摘要
多源遥感数据分析应用前需要评价影像几何校正或配准的质量以确保满足应用的需求。传统的均方根误差(RMSE)位置精度评价方法固然简单,然而无法描述校正后控制点(GCP)残差的分布特性,因而具有一定局限性。采用交叉验证法并从GCP残差角度出发,分别引入了Moran’sI空间自相关系数和标准偏差椭圆用于评价GCP残差的随机性和方向性。模拟实验结果表明,Mo-ran’sI和标准偏差椭圆可用于定量衡量GCP残差的空间随机性和分布方向性,从而更深入地分析几何校正的效果,指导选择恰当的校正模型,提高影像校正的精度。
Prior to multiple-sources remote sensing data analysis and applications,a quality assessment of geometric correction or registration is needed.Traditional position accuracy assessment is root-mean-square error（RMSE）,which is easy to calculate.However,it can not describe the spatial distribution characteristics of GCP residuals after geometric correction or registration.In this paper,cross validation is employed here to overcome traditional RMSE limitations.Furthermore,to assess the spatial characteristics of GCP residuals,Moran＇s I is adopted to test the spatial independent of GCP residuals,and standard deviational ellipse is introduced to measure the distribution trend and investigated the spatial isotropy or anisotropy for a set of GCP residuals points.Simulation experiments show that with the aid of those measurements,it can be insight into the principle of transformation function and help users to select an appropriate correction model to improve both accuracy of remote sensing image geometric correction.
出处
《遥感技术与应用》
CSCD
北大核心
2011年第2期226-232,共7页
Remote Sensing Technology and Application
基金
国家自然科学基金项目(40971222)
国家863重点项目(2006AA120106)资助
