全文下载： 
202206014.pdf
文章编号： 1672-6987（2022）06-0098-11； DOI： 10.16351/j.1672-6987.2022.06.014
刘国柱，于新(青岛科技大学 信息科学技术学院，山东 青岛 266061)
摘要： 单视图三维重建技术的研究是一个长期存在的不适定问题，随着深度学习的不断发展，基于深度学习的单视图三维重建问题得到了广泛的关注，近年来取得了显著进展。首先，从体素、点云、网格三种不同的三维模型表示形式概述了基于深度学习的单视图三维重建方法的研究进展。其次，归纳总结了目前国内外常用的三维重建数据集以及评价标准，并对一些经典方法按照不同指标进行对比分析。最后，针对三维重建在重建分辨率、重建评价指标以及现有数据集存在的问题，对未来三维重建的发展进行了展望，并探讨了三维重建技术中仍然面临的挑战及发展趋势。
关键词： 深度学习； 单视图； 三维重建； 三维模型
中图分类号： TP 301文献标志码： A
引用格式： 刘国柱,于新. 基于深度学习的单视图三维重建方法的研究进展［J］. 青岛科技大学学报(自然科学版), 2022, 43(6): 98-108.
LIU Guozhu, YU Xin. Deep learning based single view 3D reconstruction method：A survey［J］. Journal of Qingdao University of Science and Technology（Natural Science Edition), 2022， 43（6）： 98-108.
Deep Learning Based Single View 3D Reconstruction Method：A Survey
LIU Guozhu， YU Xin
(College of Information Science and Technology, Qingdao University of Science and Technology, Qingdao 266061, China)
Abstract: The research of single-view 3D reconstruction technology is a long-standing discomforting problem, and with the continuous development of deep learning, the problem of single-view 3D reconstruction based on deep learning has received extensive attention and has made significant progress in recent years. Firstly, the research progress of single-view 3D reconstruction methods based on deep learning is outlined in terms of three different 3D model representation forms, namely voxel, point cloud and mesh. Secondly, the common 3D reconstruction datasets and evaluation criteria at home and abroad are summarized, and some classical methods are compared and analyzed according to different indexes. Finally, in view of the problems of 3D reconstruction in terms of reconstruction resolution, evaluation metrics and existing datasets, the future development of 3D reconstruction is prospected, and the challenges and development trends still faced in 3D reconstruction technology are discussed.
Key words: deep learning; single view; 3D resconstruction; 3D model
收稿日期： 2021-12-08
基金项目： 山东省教研项目(重点项目) (Z2020003)；山东省教研项目(Z2016Z004，M2018X130，201801243014).
作者简介： 刘国柱(1965—), 男,教授.
