Title:
SINGLE BUILDING THREE-DIMENSIONAL RECONSTRUCTION METHOD BASED ON POINT CLOUD SEMANTIC SEGMENTATION AND STRUCTURE FITTING
Document Type and Number:
WIPO Patent Application WO/2024/077812
Kind Code:
A1
Abstract:
The present invention provides a single building three-dimensional reconstruction method based on point cloud semantic segmentation and structure fitting. The method mainly comprises: preprocessing building target point cloud data; training a building point cloud semantic segmentation network, and classifying building structure data; performing geometric classification and three-dimensional parametric fitting of a building structure; and constructing a building structure prior information topological relationship and combining single building structures. According to the present invention, when coping with the problem in the field of semantic reconstruction of a single building, the three-dimensional reconstruction, from point cloud data to semantic information, of a building structure primitive model can be implemented, and the segmentation effect, the reconstruction precision, the processing efficiency, etc. are accordingly improved.
Inventors:
SHA JIANJUN (CN)
MA CHENGYUAN (CN)
WANG XIANGWEI (CN)
WANG YANHENG (CN)
HAN MING (CN)
ZHANG XIANG (CN)
MA CHENGYUAN (CN)
WANG XIANGWEI (CN)
WANG YANHENG (CN)
HAN MING (CN)
ZHANG XIANG (CN)
Application Number:
PCT/CN2023/073652
Publication Date:
April 18, 2024
Filing Date:
January 29, 2023
Export Citation:
Assignee:
UNIV HARBIN ENG (CN)
International Classes:
G06T17/10; G06F30/13; G06N3/08
Foreign References:
CN115115839A | 2022-09-27 | |||
CN113379898A | 2021-09-10 | |||
CN114926699A | 2022-08-19 |
Other References:
CHARLES R QI: "Pointnet++: deep hierarchical feature learning on point sets in a metric space", ARXIV.ORG, CORNELL UNIVERSITY LIBRARY, ARXIV.ORG, ITHACA, 7 June 2017 (2017-06-07), Ithaca, XP093159349, Retrieved from the Internet DOI: 10.48550/arXiv.1706.02413
" Master’s Theses ", 1 May 2021, HUAZHONG NORMAL UNIVERSITY, China, article ZHANG, XIAOZHI: "Research on 3D Semantic Modeling of Building Point Cloud Based on PointNet", pages: 1 - 69, XP009553944
" Master’s Theses ", 1 May 2021, HUAZHONG NORMAL UNIVERSITY, China, article ZHANG, XIAOZHI: "Research on 3D Semantic Modeling of Building Point Cloud Based on PointNet", pages: 1 - 69, XP009553944
Attorney, Agent or Firm:
BEIJING GAOWO LAW FIRM (CN)
Download PDF:
Previous Patent: INTERMEDIATE CHIP, AND PROCESSING METHOD FOR CHIP STACKED PACKAGE
Next Patent: QUICK ACUTE-ANGLE CLAMP FOR WELDING ROBOT
Next Patent: QUICK ACUTE-ANGLE CLAMP FOR WELDING ROBOT