Multiple representation approach to achieve high-performance spatial queries of 3D BIM data using a relational database

We propose an approach using multiple representations to achieve high-performance spatial queries on 3D BIM data. The geometry data is first stored as a simple polyhedron representation inside a relational database supporting the storage of 3D geometry polyhedra. Additional representations are deriv...

Full description

Saved in:
Bibliographic Details
Published inAutomation in construction Vol. 81; pp. 369 - 388
Main Authors Solihin, Wawan, Eastman, Charles, Lee, Yong-Cheol
Format Journal Article
LanguageEnglish
Published Amsterdam Elsevier B.V 01.09.2017
Elsevier BV
Subjects
3D spatial query
Approximation
BIM
Boundary representation
Boxes
Building information modeling
Decomposition
Geometry
Indexing
Mathematical analysis
Polyhedra
Queries
Query languages
Relational data bases
Relational database
Structured Query Language-SQL
Studies
Geometry
BIM
Relational database
3D spatial query
Online AccessGet full text

Cover

More Information
Summary:We propose an approach using multiple representations to achieve high-performance spatial queries on 3D BIM data. The geometry data is first stored as a simple polyhedron representation inside a relational database supporting the storage of 3D geometry polyhedra. Additional representations are derived from the main geometry for ready use in many applications such as rule checking. The additional representations include a shape approximation using bounding boxes as the roughest approximation, octree decomposition using non-uniform octree cell size that also serves as a spatial index for the geometry, and its boundary representations by the topological faces that are indexed. The octree decomposition is used for efficient spatial indexing. The octree cells are decoded into a one-dimensional string so that it can be indexed using the ordinary binary tree index supported in the relational database, allowing for high-performance spatial queries. Commonly used spatial queries such as intersection and spatial relation between many objects can be efficiently performed using ordinary SQL queries. Using the proposed method, high performance integrated queries of the complete BIM data that includes its geometry can be achieved. The performance of this approach is validated by comparing identical queries against the same data using the native database supported R-tree indexing and the octree-based indexing. It shows a significant performance improvement that is 8 to 234 times faster and the performance is noticeably more consistent. •We propose a multiple representation scheme for geometries in a database.•We describe integrated queries to a database for both geometry and other data.•High-performance spatial queries utilizing octree indexes in a standard SQL•Demonstrates the effectiveness integrating spatial operations on realistic BIM data.•Superior spatial query performance using our approach compared to Oracle Spatial
ISSN:0926-5805
1872-7891
DOI:10.1016/j.autcon.2017.03.014