基于深度学习的三维模型检索算法综述
作者:
作者单位:

天津大学电气自动化与信息工程学院,天津300072

作者简介:

通讯作者:

基金项目:

国家自然科学基金(61772359,61902277)资助项目;天津市新一代人工智能重大专项 (19ZXZNGX00110,18ZXZNGX00150)资助项目;中国博士后科学基金(2020M680884)资助项目。


Review of 3D Model Retrieval Algorithms Based on Deep Learning
Author:
Affiliation:

School of Electrical and Information Engineering, Tianjin University, Tianjin 300072, China

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
    摘要:

    近年来,深度学习被广泛应用于各个领域并取得了显著的进展,如何利用深度学习高效管理呈爆炸式增长的三维模型一直是一个研究热点。本文介绍了发展至今主流的基于深度学习的三维模型检索算法,并根据实验得出的算法性能评估分析了其优缺点。根据检索任务的不同,可将主要的三维模型检索算法分为两类:(1)基于模型的三维模型检索方法,即检索对象与被检索对象都是三维模型,按照对三维模型的表示方式不同,可进一步分为基于体素、基于点云和基于视图的方法;(2)基于二维图像的跨域三维模型检索方法,即检索对象是二维图像,被检索对象是三维模型,包括基于二维真实图像和基于二维草图的三维模型检索方法。最后,对基于深度学习的三维模型检索算法目前存在的问题进行分析和讨论,并展望未来发展的新方向。

    Abstract:

    In recent years, deep learning has been widely used and achieved significant development in various fields. How to utilize deep learning to effectively manage the explosive increasing 3D models becomes a hot topic. This paper introduces the mainstream algorithms for deep learning based 3D model retrieval and analyzes the advantages and disadvantages according to the experimental performance. In terms of the retrieval tasks, 3D model retrieval algorithms are classified into two categories: (1) Model-based 3D model retrieval algorithms require that both query and gallery are 3D models. It can be further divided into voxel-based method, point cloud-based method and view-based method in regard of different representations of 3D models. (2) For 2D image-based cross-domain 3D model retrieval algorithms, the query is 2D image while the gallery is 3D model. It can be classified to 2D real image-based method and 2D sketch-based method. Finally, we analyze and discuss existing issues of deep learning based 3D model retrieval methods, and predict possible promising directions for this research topic.

    表 2 基于模型的三维模型检索算法在ModelNet40和ShapeNet Core55数据集上的性能比较Table 2 Performance comparison of the model-based 3D model retrieval algorithms on ModelNet40 and ShapeNet Core55 datasets
    表 1 三维模型检索算法对比Table 1 Comparison of 3D model retrieval algorithms
    表 4 基于二维草图的三维模检索算法在SHREC’14 LSSTB数据集上性能比较Table 4 Performance comparison of the 2D sketch-based 3D model retrieval algorithms on SHREC’14 LSSTB dataset
    图1 3D ShapeNets结构示意图Fig.1 Schematic of the 3D ShapeNets structure
    图2 DLAN结构示意图Fig.2 Schematic of DLAN structure
    图3 PointNet结构示意图Fig.3 Schematic of the PointNet structure
    图4 PointNet++结构示意图Fig.4 Schematic of the PointNet++ structure
    图5 基于视图的三维模型检索算法结构示意图Fig.5 Schematic of the view-based 3D model retrieval algorithm
    图6 基于二维真实图像的跨域三维模型检索算法结构示意图Fig.6 Schematic of the 2D real image-based cross-domain 3D model retrieval algorithm
    图7 CDMR结构示意图[52]Fig.7 Schematic diagram of CDMR structure[52]
    图8 Siamese-CNN结构示意图Fig.8 Schematic diagram of Siamese-CNN structure
    图9 DCML结构示意图Fig.9 Schematic diagram of DCML structure
    图10 ShapeNet Core55数据集示例Fig.10 Example of the ShapeNet Core55 dataset
    图11 ModelNet40数据集示例Fig.11 Example of the ModelNet40 dataset
    图12 SHREC’14 LSSTB数据集示例Fig.12 Examples of SHREC’14 LSSTB dataset
    图13 MI3DOR数据集示例Fig.13 Examples of MI3DOR dataset
    表 3 基于二维真实图像的三维模型检索在MI3DOR数据集上的性能比较Table 3 Performance comparison of the 2D real image-based 3D model retrieval algorithms on MI3DOR dataset
    参考文献
    相似文献
    引证文献
引用本文

刘安安,李天宝,王晓雯,宋丹.基于深度学习的三维模型检索算法综述[J].数据采集与处理,2021,36(1):1-21

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
历史
  • 收稿日期:2020-12-14
  • 最后修改日期:2021-01-04
  • 录用日期:
  • 在线发布日期: 2021-02-01