摘要
三维扫描与建模是一种能够获取目标场景三维信息的重要立体表示方法,在观演空间舞台效果展示、虚拟看房、文物保护等领域具有广泛应用。本文主要研究了一种三维扫描与建模方法,包括三维深度感知、点云配准与三维点云建模等内容。首先提出了全向环结构光深度感知方法;然后建立了混合点云配准模型,粗配准采用穷举搜索与特征匹配算法,精配准采用基于双向匹配与比率筛选的ICP(Iterative Closest Point,ICP)配准算法;最后通过Geomagic等软件对点云进行主动建模。实验结果表明,本文方法可以获得高精度、高鲁棒性的大场景三维扫描与建模结果。
3D scanning and modeling is an important three-dimensional representation method that can obtain the three-dimensional information of the target scene.It is widely used in the fields of performance space stage effect display,virtual house viewing,cultural relics protection and so on.This paper mainly studies a 3D scanning and modeling method,including 3D depth perception,point cloud registration and 3D point cloud modeling.Firstly,an omni-directional ring structured light depth sensing method is proposed.Then a hybrid point cloud registration model is established.The coarse registration adopts exhaustive search and feature matching algorithm,and the fine registration adopts iterative closest point(ICP)registration algorithm based on two-way matching and ratio screening.Finally,the point cloud is actively modeled by Geomagic and other software.Experimental results show that this method can obtain high-precision and robust 3D point cloud scanning and modeling results of large scenes.
作者
李文浩
贾同
吕朝辉
孙小钧
黄俊文
张松娜
LI Wenhao;JIA Tong;LV Zhaohui;SUN Xiaojun;HUANG Junwen;ZHANG Songna(Northeastern University,Shenyang 110819,China;Communication University of China,Beijing 100024,China)
出处
《中国传媒大学学报(自然科学版)》
2022年第2期79-88,共10页
Journal of Communication University of China:Science and Technology
基金
观演空间静态感知技术与系统研发(2018YFB1404101)
国家重点研发计划(20200197)
自适应环境结构光生成方法研究(62173083)
国家自然科学基金(20210379)。
关键词
深度感知
点云配准
三维建模
结构光
depth perception
point cloud registration
3D modeling
structured light