摘要
当前三维重建生成密集点云的过程中存在质量评估研究并不深入的问题,而且也未形成一个完整的体系,并且缺乏对新设计软件系统的比较和验证,所以极易导致用户在选择时存在一定程度的盲目性。为了解决上述问题,对基于影像生成的点云进行深入的研究,提出通过扩展评价指标来构建点云质量评价体系的解决方案,利用数码相机和无人机获取多视影像并使用九种开源算法和商业软件生成密集点云,计算激光扫描点云与密集点云数据之间的最邻近距离。不仅从点云精度和完整度,还从点云数目、点云密度、点云生成时间、程序运行占用的内存等角度对密集点云进行评估。实验表明,研究提高了质量评估的可靠性与实用性,进一步完善了密集点云的评价体系,可为相关工程实践应用提供参考依据。
Aiming at the problem that the quality evaluation of dense point clouds generated in 3D reconstruction is not in-depth and not yet formed a complete system,and the lack of comparison and verification of new design software system,so it is easy to lead to blindness when users choose.In order to solve the above problems,this paper makes an in-depth study on the point cloud based on image generation,and proposes a solution to build the quality evaluation system of point cloud by expanding the evaluation index.The multi view images are obtained by digital camera and UAV,and use nine open source software and commercial software to generate dense point clouds.The nearest distance between laser scanning point cloud and dense point cloud data is calculated.Not only from the point cloud accuracy and integrity,but also from the point cloud number,point cloud density,point cloud generation time,the memory occupied by program operation,etc.The experiments show that this study improves the reliability and practicability of quality assessment,and further improves the evaluation system of dense point cloud,which can provide reference for the application of related engineering practice.
作者
李彩林
孙延坤
王佳文
苏本娅
朱夫宏
杨长磊
LI Cailin;SUN Yankun;WANG Jiawen;SU Benya;ZHU Fuhong;YANG Changlei(Institute of Architecture and Engineering,Shandong University of Technology,Zibo Shandong 255000,China)
出处
《激光杂志》
CAS
北大核心
2021年第11期79-86,共8页
Laser Journal
基金
国家自然科学基金(No.41601496,41701525)。
关键词
多视影像
密集匹配
点云
质量评估
multi-view image
dense matching
point cloud
quality evaluation
作者简介
李彩林(1985-),男,博士,副教授,研究方向为计算机视觉与摄影测量。E-mail:licailin@sdut.edu.cn;通信作者:孙延坤,男,硕士研究生,研究方向为计算机视觉与摄影测量。E-mail:syk-sdut@foxmail.com。
