Open AccessA Pose Transformation Estimation Method Based on Feature Extraction and Matching in Ceramic Green Body Bonding Process
Author(s) -
Qihang Ma,
Jian Zhang
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1965/1/012002
Subject(s) - point cloud , iterative closest point , transformation (genetics) , ransac , pose , matching (statistics) , artificial intelligence , computer science , histogram , green body , feature (linguistics) , computer vision , feature extraction , process (computing) , rigid transformation , transformation matrix , pattern recognition (psychology) , ceramic , mathematics , image (mathematics) , materials science , statistics , biochemistry , chemistry , linguistics , philosophy , kinematics , composite material , classical mechanics , physics , gene , operating system
Bonding the ceramic green-bodies of the seat ring to the cup body is an important step in the production process of the toilet. This paper proposes a pose transformation estimation method based on feature extraction and matching in ceramic green body bonding process. This method uses a Local Feature Statistics Histogram (LFSH) to extract the features of the bonding surfaces and find the corresponding point sets. Then, a Sample Consensus Initial Aligment (SAC-IA) algorithm combined with LFSH features is used to perform coarse registration calculation. Finally, using the Iterative Cloest Point (ICP) algorithm, the result of the coarse registration is used as the initial value to complete the fine registration and obtain the pose transformation matrix. Before registration, the relative poses of the two point clouds are randomly transformed. This method is tested on the point cloud data of the cup body and the seat ring. Experimental results show that the role of SAC-IA can increase the speed of registration compared to using ICP alone.