Deep Correspondence Matching Based Robust Point Cloud Registration of Profiled Parts

被引:7
作者
Peng, Weixing [1 ]
Wang, Yaonan [1 ]
Zhang, Hui [2 ]
Cao, Yihong [3 ]
Zhao, Jiawen [1 ]
Jiang, Yiming [2 ]
机构
[1] Hunan Univ, Coll Elect & Informat Engn, Changsha 410082, Peoples R China
[2] Hunan Univ, Coll Robot, Changsha, Peoples R China
[3] Hunan Univ, Coll Comp Sci & Elect Engn, Changsha, Peoples R China
基金
中国国家自然科学基金;
关键词
Profiled parts; point cloud; coordinate alignment; deep neural network; automatic manufacturing;
D O I
10.1109/TII.2023.3287074
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Due to ability to estimate the spatial transformation of coordinate frames, point cloud registration is a fundamental technique in manufacturing. Previous methods prone to converge to wrong local minima, in the cases of large initialization, noise, outliers, and partiality. This study presents a new learning-based robust point cloud registration approach to predict a rigid transformation in a one-shot way. Our network aims to determine a matchability matrix to yield an accurate registration result. Each element of the matchability matrix refers to similarity of learned per-point embeddings and represents the probability of a potential correspondence. Two major blocks are developed to guide the matchability matrix to represent correct correspondences: 1) an attention block is introduced to enhance the discriminativeness of learned per-point embeddings; 2) a zero-mean Gaussian based annealing layer and a differentiable Sinkhorn normalization layer are designed to enforce a permutation matchability matrix. With the matchability matrix, an intuitive solution is integrated to obtain the relative transformation of the source and target point clouds. Different from existing work, our network can handle partially overlapped point-cloud pairs effectively. Experimental results demonstrate the superiority of the proposed approach over the state-of-the-art registration approaches in terms of accuracy and robustness.
引用
收藏
页码:2129 / 2143
页数:15
相关论文
共 50 条
[41]   CONTOURS BASED APPROACH FOR THERMAL IMAGE AND TERRESTRIAL POINT CLOUD REGISTRATION [J].
Bennis, Abdelhamid ;
Bombardier, Vincent ;
Thiriet, Philippe ;
Brie, David .
XXIV INTERNATIONAL CIPA SYMPOSIUM, 2013, 40-5-W2 :97-101
[42]   PointpartNet: 3D point-cloud registration via deep part-based feature extraction [J].
Yan, Shixun ;
Pathak, Sarthak ;
Umeda, Kazunori .
ADVANCED ROBOTICS, 2022, 36 (15) :724-734
[43]   Single Match: a point cloud coarse registration method with single match point and deep-learning describer [J].
Wu, Rui ;
Nie, Jianhui ;
Gao, Hao ;
Liu, Ye ;
Lu, Haotian .
MULTIMEDIA TOOLS AND APPLICATIONS, 2022, 81 (12) :16967-16986
[44]   Point Cloud based Hierarchical Deep Odometry Estimation [J].
Nowruzi, Farzan Erlik ;
Kolhatkar, Dhanvin ;
Kapoor, Prince ;
Laganiere, Robert .
PROCEEDINGS OF THE 7TH INTERNATIONAL CONFERENCE ON VEHICLE TECHNOLOGY AND INTELLIGENT TRANSPORT SYSTEMS (VEHITS), 2021, :112-121
[45]   SAP-Net: A Simple and Robust 3D Point Cloud Registration Network Based on Local Shape Features [J].
Li, Jinlong ;
Li, Yuntao ;
Long, Jiang ;
Zhang, Yu ;
Gao, Xiaorong .
SENSORS, 2021, 21 (21)
[46]   Point cloud based hand gesture recognition using template matching [J].
Zheng, Lingfei ;
Hu, Zhubing ;
Yao, Meiling ;
Xu, Pengwei ;
Ma, Jing .
JOURNAL OF INTELLIGENT & FUZZY SYSTEMS, 2024, 46 (01) :2615-2627
[47]   Learning Representative Features by Deep Attention Network for 3D Point Cloud Registration [J].
Xia, Xiaokai ;
Fan, Zhiqiang ;
Xiao, Gang ;
Chen, Fangyue ;
Liu, Yu ;
Hu, Yiheng .
SENSORS, 2023, 23 (08)
[48]   Triple screening point cloud registration method based on image and geometric features [J].
Zhong, Linshan ;
Ying, Jie ;
Yang, Haima ;
Jin, Liu .
OPTIK, 2021, 246
[49]   Three-Dimensional Point Cloud Registration Based on Normal Vector Angle [J].
Liang Li ;
Xingyan Cao ;
Jie Sun .
Journal of the Indian Society of Remote Sensing, 2019, 47 :585-593
[50]   AUTOMATIC FEATURE-BASED POINT CLOUD REGISTRATION FOR A MOVING SENSOR PLATFORM [J].
Weinmann, Martin ;
Dittrich, Andre ;
Hinz, Stefan ;
Jutzi, Boris .
ISPRS HANNOVER WORKSHOP 2013, 2013, 40-1 (W-1) :373-378