Login
- Dokumenttyp:
- Bachelor's thesis
- Autor(en):
- Thiele, C.
- Titel:
- Matching construction drawings to Building Information Models using 2D shape distribution
- Abstract:
- Building Information Modeling (BIM) is on the verge of being incorporated into the construction industry. Even though the advantages of using BIM are apparent, the industry is still mainly based on two dimensional (2D) construction drawings. BIM application has therefore been adapted to derive those drawings from the three dimensional (3D) model. Due to the high amount of parties involved in the building process, there was a need for a vendor-neutral format to accomplish this task. The disadvantage of such a format is the fact that the model and the drawings become disjointed once extraction is completed and, therefore, changes applied to one will not be adjusted automatically on the other. The discrepancies that arise during this process are the source of various possible errors which could be eliminated if the drawings were aligned to each other regularly to ensure their conformity. The automation of this process, also called the registration problem, has faced challenges in the past and an adequate solution for construction drawings is yet to be found. This thesis focuses on the second step of the registration problem, the similarity measurement and matching, using the concept of 2D shape histograms as proposed by (Pu & Ramani, 2006) in their paper "On visual similarity based 2D drawing retrieval". The method delivered promising results when the drawings had similar contours with different interior features, yet as it does not support partial matching, it proves to be impractical for the purpose of this thesis. However, two methods that could potentially solve this issue are proposed as possible future solutions. One is the combination of this method with 2.5D spherical harmonics. The other consists in trying to overlap two cross-sections of the same drawing at different heights to bypass the missing capability of partial matching observed on this tested descriptor.
- Stichworte:
- BIM; LOCenter; IFC;
- Fachgebiet:
- ALL Allgemeines
- Aufgabensteller:
- Trzeciak, M.
- Jahr:
- 2019
- Jahr / Monat:
- 2019-04
- Monat:
- Apr
- Hochschule / Universität:
- Technische Universität München
BibTeX