基于 Kriging 方法的动态地形反向多级细化研究

摘要

针对层次细节模型(LOD)只能实现数据精度低于原始 DEM 数据的动态地形建模过程,提出了基于 Kriging 方法的动态地形反向细化方法(Levels Of Backward Detail,LOBD)。该方法提取出需反向细化区域,判断多级细化级别,分析区域的空间属性相关性,根据最优条件拟合变异函数,根据 Kriging 方法对目标区域进行插值,生成精度高于原始 DEM 数据精度的区域数据。设计了按独立树结构的插值数据动态存储方式。建模实验表明,新的动态建模方法的细节层次可以比 LOD 方法至少增加两级,并且保持渲染效率基本不变。%Aiming at the limitation that the Level Of Detail(LOD)model can only achieve lower data resolution precision than the original Digital Elevation Model(DEM)data within dynamic terrain modeling process, dynamic terrain Levels Of Backward Detail(LOBD)method based on Kriging algorithm is proposed. This method extracts the area to be reversely refined, deter-mines the multi-level refinement rank, analyzes the region spatial properties interdependency, and fits the variance function according to optimal conditions, interpolates data to the target area according to Kriging interpolation algorithm, and then gener-ates higher precision data than the original DEM data. The interpolation data dynamic store method according to the independent tree structure is also designed. Modeling experiments show that this new dynamic modeling method can increase at least two detail levels than LOD method, and the efficiency of rendering is kept.