基于置信区间估计理论的改进型移动曲面算法

Improved Moving Surface Algorithm Based on Confidence Interval Estimation Theory

经典移动曲面滤波算法可以对多种地形取得较好的滤波效果,但是该算法取格网内最低点作为地面点时仍存在粗差问题。基于此,提出一种置信区间检验方法,利用残差、均方根误差和置信概率作为参考值,选择最佳的初始种子点。采用格网重叠方式解决相邻格网间板块化问题,同时应用分层聚类自适应阈值确定方法确定高差阈值。对特殊种子点不足、难以拟合曲面的特殊格网,建立平面,设置阈值判断平面是否满足条件。对改进型移动曲面滤波算法与经典移动曲面滤波算法进行定性和定量实验对比,结果证明改进型移动曲面算法对粗差地形可以获得更佳的滤波效果。

The classical moving surface filtering algorithm demonstrates a significant effect of filtering on various terrains. However, the gross errors still exist when the lowest point of the grid is used as the ground point in the moving surface grid algorithm. This study proposes a confidence interval test method to select the best initial seed points by using the residual, mean square error, and confidence probability as the reference values. The grid overlap method is used to solve the problem of the fracture layer between adjacent grids. Moreover, the hierarchical clustering adaptive threshold determination method is used to determine the elevation difference threshold. For the condition where special grids with insufficient special seed points cannot fit the surface, planes are established and thresholds are set to determine whether the plane can meet the condition. This study compares the improved moving surface filtering algorithm with the classical moving surface filtering algorithm by qualitative and quantitative experiments. The findings demonstrate that the improved moving surface algorithm can obtain a better filtering effect than the classical moving surface algorithm.