Learning local shape descriptors for computing non-rigid dense correspondence

被引:0
|
作者
Jianwei Guo [1 ]
Hanyu Wang [2 ]
Zhanglin Cheng [3 ]
Xiaopeng Zhang [1 ]
Dong-Ming Yan [1 ]
机构
[1] National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences
[2] University of Maryland-College Park
[3] Shenzhen Visu CA Key Lab, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences
基金
中国国家自然科学基金; 国家重点研发计划; 北京市自然科学基金;
关键词
local feature descriptor; triplet CNN; dense correspondence; geometry image; non-rigid shape;
D O I
暂无
中图分类号
TP391.41 []; TP18 [人工智能理论];
学科分类号
080203 ; 081104 ; 0812 ; 0835 ; 1405 ;
摘要
A discriminative local shape descriptor plays an important role in various applications. In this paper, we present a novel deep learning framework that derives discriminative local descriptors for deformable 3D shapes. We use local "geometry images" to encode the multi-scale local features of a point, via an intrinsic parameterization method based on geodesic polar coordinates. This new parameterization provides robust geometry images even for badly-shaped triangular meshes. Then a triplet network with shared architecture and parameters is used to perform deep metric learning;its aim is to distinguish between similar and dissimilar pairs of points. Additionally, a newly designed triplet loss function is minimized for improved, accurate training of the triplet network. To solve the dense correspondence problem, an efficient sampling approach is utilized to achieve a good compromise between training performance and descriptor quality. During testing,given a geometry image of a point of interest, our network outputs a discriminative local descriptor for it.Extensive testing of non-rigid dense shape matching on a variety of benchmarks demonstrates the superiority of the proposed descriptors over the state-of-the-art alternatives.
引用
收藏
页码:95 / 112
页数:18
相关论文
共 50 条
  • [31] Find dense correspondence between high resolution non-rigid 3D human faces
    Liu, Jian
    Zhang, Quan
    Tang, Chaojing
    Communications in Computer and Information Science, 2014, 472 : 259 - 275
  • [32] Non-rigid curve correspondence for estimating heart motion
    Tagare, HD
    INFORMATION PROCESSING IN MEDICAL IMAGING, 1997, 1230 : 489 - 494
  • [33] Find Dense Correspondence between High Resolution Non-rigid 3D Human Faces
    Liu, Jian
    Zhang, Quan
    Tang, Chaojing
    BIO-INSPIRED COMPUTING - THEORIES AND APPLICATIONS, BIC-TA 2014, 2014, 472 : 259 - 275
  • [34] Rigid and Non-rigid Shape Matching for Mechanical Components Retrieval
    Albarelli, Andrea
    Bergamasco, Filippo
    Torsello, Andrea
    COMPUTER INFORMATION SYSTEMS AND INDUSTRIAL MANAGEMENT (CISIM), 2012, 7564 : 168 - 179
  • [35] A Condition Number for Non-Rigid Shape Matching
    Ovsjanikov, Maks
    Huang, Qi-Xing
    Guibas, Leonidas
    COMPUTER GRAPHICS FORUM, 2011, 30 (05) : 1503 - 1512
  • [36] Example based non-rigid shape detection
    Zheng, Yefeng
    Zhou, Xiang Sean
    Georgescu, Bogdan
    Zhou, Shaohua Kevin
    Comaniciu, Dorin
    COMPUTER VISION - ECCV 2006, PT 4, PROCEEDINGS, 2006, 3954 : 423 - 436
  • [37] Automatic generation of dense non-rigid optical flow
    Le, Hoang-An
    Nimbhorkar, Tushar
    Mensink, Thomas
    Baslamisli, Anil S.
    Karaoglu, Sezer
    Gevers, Theo
    COMPUTER VISION AND IMAGE UNDERSTANDING, 2021, 212
  • [38] NON-RIGID 3D SHAPE RECOGNITION VIA DICTIONARY LEARNING
    Zhou, Yin
    Liu, Kai
    Barner, Kenneth E.
    2013 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP), 2013, : 3502 - 3506
  • [39] Learning to Optimize Non-Rigid Tracking
    Li, Yang
    Bozic, Aljaz
    Zhang, Tianwei
    Ji, Yanli
    Harada, Tatsuya
    Niessner, Matthias
    2020 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2020, : 4909 - 4917
  • [40] Learning Dense Correspondences for Non-Rigid Point Clouds With Two-Stage Regression
    Wang, Kangkan
    Zhang, Guofeng
    Zheng, Huayu
    Yang, Jian
    IEEE TRANSACTIONS ON IMAGE PROCESSING, 2021, 30 : 8468 - 8482