论文部分内容阅读
激光雷达数据滤波是为了获取高精度数字高程模型而改进的多级移动曲面拟合滤波方法。预处理剔除了粗差点和大部分建筑物侧面回波点,再通过格网化分,建立数据索引,相邻曲面保持一定的重叠度,利用最小二乘法求解拟合曲面参数,自动设置阈值后进行滤波。最后,利用多尺度加权插值方法生成数字高程模型。采用国际摄影测量与遥感学会(ISPRS)提供的测试数据进行实验。结果表明:改进的算法降低了三类误差产生的可能,第一实验区的总误差由8.39%降低到7.40%,第二实验区的总误差由6.39%降低到5.86%,生成的数字高程模型满足精度要求,具有推广性。
Lidar data filtering is an improved multi-level moving surface fitting filtering method for obtaining high-precision digital elevation model. The preprocessing eliminates the coarse points and most building side echo points, and then establishes the data index through grid points. The adjacent surfaces maintain a certain degree of overlap. The least square method is used to solve the fitting surface parameters and the threshold value is automatically set Filtering. Finally, the digital elevation model is generated by using the multi-scale weighted interpolation method. Experiments were conducted using the test data provided by the International Society for Photogrammetry and Remote Sensing (ISPRS). The results show that the improved algorithm can reduce the possibility of three types of errors. The total error of the first experimental zone is reduced from 8.39% to 7.40%, and the total error of the second experimental zone is reduced from 6.39% to 5.86%. The generated digital elevation model Meet the accuracy requirements, with promotion.