为提高城市区LIDAR数据滤波精度, 提出了一种多级移动曲面拟合滤波方法。建立区块网格搜寻及索引机制完成对离散LIDAR点云的标示; 通过建立二次多项式完成参考曲面的拟合, 不同窗口大小获得不同层次的拟合曲面; 设置自适应阈值, 完成地面点与非地面点的判断。精度评价结果表明, 该滤波算法误差在1m以内, 能够满足实际应用的需求。

LIDAR data is accurate 3D data of terrain acquired from airborne laser detection and ranging system. Compared with hardware technique, data post-processing technique of LIDAR data is weak and time consumed. In this situation, a HMCFA (Hierarchical Moving Curved Fitting Algorithm) filtering method of LIDAR data is reported in this paper. Firstly, block grid searching and indexing mechanism is set up to label discrete LIDAR cloud points. Secondly, quadratic polynomial is set up to fit land terrain with different window size. At last, adaptive threshold is used to distinguish ground points and non-ground points. Accuracy assessment results indicate that the filter error is less than 1m, which can be used in application.