基于像素模式的海量点云可视化LOD调度方法

LOD Scheduling Method of Mass Point Cloud Visualization Based on the Pixel Model

针对受限于计算机内存,海量三维点云数据无法一次性加载显示的问题,提出了一种基于像素模式的海量点云可视化LOD调度方法。该方法主要包括生成八叉树点云索引文件和设置LOD模型调度范围两个阶段,首先采用八叉树的原理对点云数据进行分层和抽稀,生成索引文件;再基于八叉树点云索引文件构建点云LOD模型;最后根据原始点云的分辨率自动计算LOD模型节点的像素阈值。在实时可视化的过程中,以LOD模型节点在屏幕上的像素大小是否达到预定阈值为判断条件,对其进行调度,避免了距离模式下海量点云可视化出现的卡顿问题。

Due to the limitation of computer memory,the mass point cloud cannot be loaded at once.On the basis of this,we proposed a LOD scheduling method of mass point cloud visualization based on the pixel model,which consisted of two stages,namely the generation of octree and the setting pixel threshold of LOD.Firstly,we stratified and vacuated the point cloud by octree theory,and generated the indexfiles.Then,we built LOD based on the indexfiles.Finally,the pixel threshold is calculated automatically according to the resolution of point cloud.In the process of real-time visualization,taking whether the pixel size of the LOD model node on the screen reaches the predetermined threshold as the judgment condition,we scheduled LOD,which avoided the stagnation problem of mass point cloud visualization in distance mode.