面向深度学习的三维点云补全算法综述  

A survey on point cloud completion algorithms for deep learning

作　　者：胡伏原[1,2,3] 李晨露 周涛 程洪福 顾敏明 Hu Fuyuan;Li Chenlu;Zhou Tao;Cheng Hongfu;Gu Minming(School of Electronic and Information Engineering,Suzhou University of Science and Technology,Suzhou 215009,China;Jiangsu Industrial Intelligent and Low-carbon Technology Engineering Center,Suzhou 215009,China;Suzhou Key Laboratory of Intelligent and Low-carbon Technology Application,Suzhou 215009,China;School of Computer Science and Engineering,North Minzu University,Yinchuan 750021,China;Suzhou Humble Administrator’s Garden Administration Office(Suzhou Garden Museum),Suzhou 215001,China)

机构地区：[1]苏州科技大学电子与信息工程学院,苏州215009 [2]江苏省工业智能低碳技术工程中心,苏州215009 [3]苏州市智能低碳技术应用重点实验室,苏州215009 [4]北方民族大学计算机科学与工程学院,银川750021 [5]苏州市拙政园管理处(苏州市园林博物馆),苏州215001

出　　处：《中国图象图形学报》2025年第2期309-333,共25页Journal of Image and Graphics

基　　金：国家重点研发计划资助(2023YFE0116300);国家自然科学基金项目(61876121);苏州市科技发展计划项目(SS202133)。

摘　　要：点云因其丰富的信息表达能力已成为三维视觉的主要表现形式,然而实际采集到的点云数据往往因各种因素导致稀疏或残缺,严重影响点云后续处理。点云补全算法旨在从残缺点云数据中重建完整点云模型,是3D重建、目标检测和形状分类等领域的重要研究基础。目前,基于深度学习的点云补全算法逐渐成为三维点云领域的研究热点,但补全任务中模型结构、精度和效率等挑战正阻碍点云补全算法的发展。本文对深度学习背景下的点云补全算法进行系统综述,首先根据网络输入模态将点云补全算法分为两大类,即基于单模态的方法以及基于多模态的方法。接着根据三维数据表征方式将基于单模态的方法分为三大类,即基于体素的方法、基于视图的方法以及基于点的方法,并对经典方法和最新方法进行系统的分析和总结,同时结合热点模型,如生成对抗网络(generative adversarial network,GAN)、Transformer模型等进一步分类对比,评述各类模型下点云补全算法的方法特点与网络性能。再对基于多模态的方法进行实际应用分析,结合扩散模型等方法进行算法性能对比。然后总结点云补全任务中常用的数据集及评价标准,分别以多种评价标准对比分析现有基于深度学习的点云补全算法在真实数据集与多种合成数据集上的性能表现。最后根据各分类的优缺点提出点云补全算法在深度学习领域的未来发展和研究趋势,为三维视觉领域的补全算法研究者提供重要参考价值。Point clouds have become the main form of 3D vision because of their rich information expression ability.How⁃ever,the actual collected point cloud data are often sparse or incomplete due to the characteristics of the measured object,the performance of the measuring instrument,and environmental and human factors,which seriously affect the subsequent processing of the point cloud.The point cloud completion algorithm aims to reconstruct a complete point cloud model from incomplete point cloud data,which is an important research basis for 3D reconstruction,object detection,and shape classi⁃fication.With the rapid development of deep learning methods,their efficient feature extraction ability and excellent data processing ability have led them to be widely used in 3D point cloud algorithms.At present,point cloud completion algo⁃rithms based on deep learning have gradually become a research hotspot in the field of 3D point clouds.However,chal⁃lenges such as model structure,accuracy,and efficiency in completion tasks hinder the development of point cloud comple⁃tion algorithms.Examples include the problems of missing key structural information,fine-grained reconstruction,and inefficiency of the algorithm model.This study systematically reviews point cloud completion algorithms in the context of deep learning.First,according to the network input modality,the point cloud completion algorithms are divided into two categories,namely,single-modality-based methods and multimodality-based methods.Then,according to the representa⁃tion of 3D data,the methods based on a single modality are divided into three categories,namely,voxel-based methods,view-based methods,and point-based methods.The classical methods and the latest methods are systematically analyzed and summarized.The method characteristics and network performance of point cloud completion algorithms under various models were reviewed.Then,practical application analysis of the multimodal method is conducted,and the performance of the algorithm is compared wi

关 键 词：点云补全 体素方法 多模态方法 Transformer模型 扩散模型 

分 类 号：TP37[自动化与计算机技术—计算机系统结构]