基于置信区间估计理论的PTD算法种子点最优选择方法

Seed point optimal selection method of PTD algorithm based on confidence interval estimation theory

茂密森林覆盖山体是激光雷达数据处理中的重点和难点。渐进式不规则三角网致密化(PTD)滤波算法对多种地形取得较好的滤波效果,但是PTD算法仍然存在最佳种子点选择问题。文章基于PTD滤波算法种子点选取问题提出置信区间检验方法,利用残差,均方差和置信概率作为参考值,选择最佳的初始种子点,并且借助激光雷达的多重回波属性,利用多次回波的高程差值对植被进行初步判定。通过对置信区间-PTD滤波算法与经典PTD滤波算法进行定性和定量实验对比分析,结果证明,改进型PTD滤波算法在森林覆盖山体中获得更佳的滤波效果。

The dense forest covered mountain is the key and difficult point in the data processing of laser radar.The progressive triangulation densification,can achieve better filtering effect on multiple terrains,but PTD algorithm still has the best seed point selection problem.This paper is based on PTD filtering algorithm seed.The point selection problem is proposed by using the confidence interval test method,using the residual,mean square error and confidence probability as reference values to select the best initial seed point;and by using the multiple echo properties of the LiDAR,using the elevation difference of multiple echoes to the vegetation to make a preliminary judgment.By comparing the qualitative and quantitative experiments of the confidence interval PTD filtering algorithm with the classical PTD filtering algorithm,the result shows that the improved PTD filtering algorithm can obtain better filtering effect in the forest covered mountain.