基于深度学习的三维点云分割综述  

A review of deep learning based on 3D point cloud segmentation

作　　者：卢健[1] 贾旭瑞 周健[1] 刘薇[1] 张凯兵[1] 庞菲菲 LU Jian;JIA Xu-rui;ZHOU Jian;LIU Wei;ZHANG Kai-bing;PANG Fei-fei(School of Electronics and Information,Xi’an Polytechnic University,Xi’an 710600,China)

机构地区：[1]西安工程大学电子信息学院,西安710600

出　　处：《控制与决策》2023年第3期595-611,共17页Control and Decision

基　　金：国家自然科学基金项目(61971339,61471161);陕西省自然科学基金重点项目(2018JZ6002)。

摘　　要：作为三维场景理解的重要技术之一,三维点云分割受到广泛的关注,具有重要的研究价值和广阔的应用前景.基于此,梳理基于深度学习的三维点云分割技术的最新研究进展;在介绍三维点云分割常用的8个室内和室外数据集的基础上,重点阐述和分析现有主要基于深度学习的语义分割、实例分割和部件分割方法,并基于量化数据进行部分方法的效能比较;最后从10个方面总结现有方法的不足,并针对性地提出工作展望.3D point cloud segmentation, as one of the important technology of 3D scene understanding, has aroused people’s widespread interest. And it has important research value and broad application prospect. The latest research progress of 3D point cloud segmentation technology based on deep learning is sorted out. Firstly, eight indoor and outdoor common datasets frequently utilized in 3D point cloud segmentation are introduced. Then, the existing semantic segmentation, instance segmntation and part segmentation mainly via deep learning are explained and analyzed in depth,and the effectiveness of some methods is compared baesd on quantitative data. Finally, we summarize the shortcomings of the existing methods from ten aspects, and put forward the work prospects pertinently.

关 键 词：深度学习 三维点云 数据集 语义分割 实例分割 部件分割 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]