DEM粗差权值衰减迭代探测算法研究

Research on DEM Gross Error Weight Decay Iterative Detection Algorithm

为剔除经滤波处理后的机载激光雷达(LIDAR)地面点云数据中的残留非地面点,提高由其制作生产的数字高程模型(DEM)的精度,提出一种基于LIDAR点云数据所生成的不规则分布数据DEM粗差权值衰减迭代探测算法。该算法将滤波后残留的非地面点视为DEM粗差,基于地理学局部地形相似性原理,根据数据点密度及地形起伏变化程度确定局部窗口并进行二次曲面拟合,求解局部窗口内各数据点高程残差值,从而构建粗差相关检验量来定位粗差点。同时,该算法借鉴选权迭代粗差定位法思想,通过迭代,不断衰减含粗差观测点的权值,从而不断减小其对局部窗口二次曲面拟合产生的不良影响,降低地面点被误判为粗差点的概率。通过实验验证,该算法能有效提升DEM粗差探测率并极大降低粗差误判概率,粗差剔除率为98.02%,粗差误判率为1.71%。

In order to study the way to eliminate the residual non-ground points in the filtered airborne LiDAR ground point cloud data and improve the accuracy of the produced digital elevation model(DEM),this paper proposed an iterative detection algorithm based on the attenuation of DEM gross error weights for the irregularly distributed data generated from the LiDAR point cloud data.The algorithm considered the residual non-ground points as DEM gross error.Based on the principle of local terrain similarity in geography,the algorithm determined the local window according to the density of data points and the degree of change of terrain and carried out quadratic surface fitting,solved the residual value of elevation of each data point in the local window,and constructed the related test to locate the gross error.At the same time,the algorithm drew on the idea of selective iterative gross error localization method,through iteration,constantly attenuating the weights of the gross error points,so as to continuously reduce the adverse effect on the local window quadratic surface fitting and the probability of ground points being misjudged as gross error.Through experimental verification,the algorithm can effectively improve the DEM gross error detection rate and greatly reduce the gross error misjudgment probability,and its gross error rejection rate reaches 98.02%,and the gross error misjudgment rate is 1.71%.