摘要
在使用最小二乘法解算卫星遥感影像的RPC参数时,如果控制点非均匀分布或模型过度参数化,其法方程系数矩阵很容易产生病态,获得的解将偏离真值,甚至得到错误的解。使用岭估计可改善法方程的状态,保证解稳定。采用岭估计方法,通过所获取的不同岭参数对SPOT和QuickBird影像进行实验,证实L曲线法是一种稳定的、有效的岭参数确定方法,可显著提高RPC参数的解算精度。
If the distribution of the control points is asymmetric or the model is over parameterized, the problem of ill conditioned normal equation easily occurs during solving the ra tional polynomial eoefficients（RPC） of satellite imagery. Traditional least squares adjust- ment can＇t get reliable solution. Ridge estimation is introduced to ameliorate the condition of the normal equation and to ensure that the solution is reliable. The basic principle of solving RPC by using ridge estimation is introduced. SPOT and QuickBird imagery are processed by different regularization techniques. The empirical results have verified that L-curve method is a reliable and valid way for choosing the ridge parameter, and it could improve the aceura cy of the solution distinctly.
出处
《武汉大学学报（信息科学版）》
EI
CSCD
北大核心
2008年第11期1130-1133,共4页
Geomatics and Information Science of Wuhan University
基金
国家973计划资助项目(2006CB701302)
国家创新研究群体科学基金资助项目(40721001)
关键词
RPC参数
岭估计
L曲线法
rational polynomial coefficients （RPC）
ridge estimation
L-curve method
