应用于增强现实变电运检三维场景重建的增量八叉树算法  

The incremental octree schedule of digital transformer substation scenario reconstructions in augmented-reality environments

作　　者：焦进 迟小羽 赖秋 杨子力 储园 杨坤莲 杨锐 JIAO Jin;CHI Xiaoyu;LAI Qiu;YANG Zili;CHU Yuan;YANG Kunlian;YANG Rui(Yunnan Power Supply Co.,Ltd.,Qujing Bureau,Qujing 655000,China;Qingdao Research Institute of BUAA,Qingdao 266100,China;Kunming Escher Technology Limited Company,Kunming 650000,China)

机构地区：[1]云南电网有限责任公司曲靖供电局,云南曲靖655000 [2]北京航空航天大学青岛研究院,山东青岛266100 [3]昆明埃舍尔科技有限公司,云南昆明650000

出　　处：《应用科技》2020年第5期58-63,共6页Applied Science and Technology

摘　　要：为解决数字增强现实(AR)眼镜设备上海量点云数据管理和三维重建的难题,本文提出了一种增量八叉树算法。该算法可以均衡地将点云数据保存到空间索引结构中,实现实时渲染过程中点云数据的动态调度,使场景点云的密度分布与视场角和视点距离线性相关,且有效降低了点云存储空间。利用Hololens增强现实眼镜对本文算法进行了测试验证,实验结果表明,利用本算法可以加载和调度十亿点云数量的变电站场景点云数据,并以不小于25 f/s的帧速率实时显示数字化场景。It is difficult for developers to consider managing huge point cloud data during the 3D reconstruction,and render them smoothly on the augmented-reality(AR)glasses device for on-site use.In this paper,we present a new incremental octree method,which saves point cloud data in a spatial index structure in a balanced way,realize dynamic scheduling of point cloud data in the realtime rendering process without producing redundant data.This method makes the density distribution of point cloud in the scene linearly related to the angle of view and the distance between viewpoints,and effectively reduce the point cloud storage space.The method is already tested in a transformer substation scenario with billion points.The digital scene is displayed on a Hololens helmet at 25 f/s,which is considered to have comparatively higher practical values in transformer substation systematic maintenance applications.

关 键 词：数字孪生 增强现实 动态调度 增量八叉树 Hololens 点云 变电运检 三维重建 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]