抗差最小二乘配置在数字化地图几何纠正中的应用讨论

Discussion on the Application of Robust Least Squares Collocation in Geometric Correction of Digital Map

抗差估计具有较好的抗拒异常观测值及粗差的能力,而最小二乘配置又能较好地处理系统误差,本文结合两者的优点,利用抗差最小二乘配置对数字化地图进行几何纠正,其中对协方差函数采用抗差拟合,得到了较好的结果。实验证明在GIS数据处理的扫描数字化地图几何纠正中,抗差最小二乘配置在抗拒异常值和处理系统误差方面优于单纯的最小二乘估计和单纯的最小二乘配置方法。

Robust estimation can resist abnormal observations and coarse error well, and least squares collocation can preferably process systemic errors. Combined with their respective advantages, this paper implemented the geometric correction for digital map using robust least squares collocation,and obtained satisfied results. The procedure in robust fitting for covariance function is also discussed. The experimental results showed that the robust least squares collocation is better than pure least squares and least squares collocation in resisting abnormal observation and processing systemic errors in the geometric correction of scanned digital map for GIS data process.