Semantic Extraction of Permanent Structures for the Reconstruction of Building Interiors from Point Clouds
The extraction of permanent structures (such as walls, floors, and ceilings) is an important step in the reconstruction of building interiors from point clouds. These permanent structures are, in general, assumed to be planar. However, point clouds from building interiors often also contain clutter...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2020-12-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/20/23/6916 |
| _version_ | 1797545794521268224 |
|---|---|
| author | Inge Coudron Steven Puttemans Toon Goedemé Patrick Vandewalle |
| author_facet | Inge Coudron Steven Puttemans Toon Goedemé Patrick Vandewalle |
| author_sort | Inge Coudron |
| collection | DOAJ |
| description | The extraction of permanent structures (such as walls, floors, and ceilings) is an important step in the reconstruction of building interiors from point clouds. These permanent structures are, in general, assumed to be planar. However, point clouds from building interiors often also contain clutter with planar surfaces such as furniture, cabinets, etc. Hence, not all planar surfaces that are extracted belong to permanent structures. This is undesirable as it can result in geometric errors in the reconstruction. Therefore, it is important that reconstruction methods can correctly detect and extract all permanent structures even in the presence of such clutter. We propose to perform semantic scene completion using deep learning, prior to the extraction of permanent structures to improve the reconstruction results. For this, we started from the ScanComplete network proposed by Dai et al. We adapted the network to use a different input representation to eliminate the need for scanning trajectory information as this is not always available. Furthermore, we optimized the architecture to make inference and training significantly faster. To further improve the results of the network, we created a more realistic dataset based on real-life scans from building interiors. The experimental results show that our approach significantly improves the extraction of the permanent structures from both synthetically generated as well as real-life point clouds, thereby improving the overall reconstruction results. |
| first_indexed | 2024-03-10T14:21:03Z |
| format | Article |
| id | doaj.art-18a21a243aad446cb65576a3a5c17ac7 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-10T14:21:03Z |
| publishDate | 2020-12-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-18a21a243aad446cb65576a3a5c17ac72023-11-20T23:22:17ZengMDPI AGSensors1424-82202020-12-012023691610.3390/s20236916Semantic Extraction of Permanent Structures for the Reconstruction of Building Interiors from Point CloudsInge Coudron0Steven Puttemans1Toon Goedemé2Patrick Vandewalle3Flanders Make, 3001 Heverlee, BelgiumFlanders Innovation & Entrepreneurship (VLAIO), 1030 Brussel, BelgiumEAVISE, PSI, Department of Electrical Engineering (ESAT), KU Leuven, Jan Pieter De Nayerlaan 5, 2860 Sint-Katelijne-Waver, BelgiumEAVISE, PSI, Department of Electrical Engineering (ESAT), KU Leuven, Jan Pieter De Nayerlaan 5, 2860 Sint-Katelijne-Waver, BelgiumThe extraction of permanent structures (such as walls, floors, and ceilings) is an important step in the reconstruction of building interiors from point clouds. These permanent structures are, in general, assumed to be planar. However, point clouds from building interiors often also contain clutter with planar surfaces such as furniture, cabinets, etc. Hence, not all planar surfaces that are extracted belong to permanent structures. This is undesirable as it can result in geometric errors in the reconstruction. Therefore, it is important that reconstruction methods can correctly detect and extract all permanent structures even in the presence of such clutter. We propose to perform semantic scene completion using deep learning, prior to the extraction of permanent structures to improve the reconstruction results. For this, we started from the ScanComplete network proposed by Dai et al. We adapted the network to use a different input representation to eliminate the need for scanning trajectory information as this is not always available. Furthermore, we optimized the architecture to make inference and training significantly faster. To further improve the results of the network, we created a more realistic dataset based on real-life scans from building interiors. The experimental results show that our approach significantly improves the extraction of the permanent structures from both synthetically generated as well as real-life point clouds, thereby improving the overall reconstruction results.https://www.mdpi.com/1424-8220/20/23/6916deep learningsemantic segmentationsemantic completionindoor 3D reconstruction |
| spellingShingle | Inge Coudron Steven Puttemans Toon Goedemé Patrick Vandewalle Semantic Extraction of Permanent Structures for the Reconstruction of Building Interiors from Point Clouds Sensors deep learning semantic segmentation semantic completion indoor 3D reconstruction |
| title | Semantic Extraction of Permanent Structures for the Reconstruction of Building Interiors from Point Clouds |
| title_full | Semantic Extraction of Permanent Structures for the Reconstruction of Building Interiors from Point Clouds |
| title_fullStr | Semantic Extraction of Permanent Structures for the Reconstruction of Building Interiors from Point Clouds |
| title_full_unstemmed | Semantic Extraction of Permanent Structures for the Reconstruction of Building Interiors from Point Clouds |
| title_short | Semantic Extraction of Permanent Structures for the Reconstruction of Building Interiors from Point Clouds |
| title_sort | semantic extraction of permanent structures for the reconstruction of building interiors from point clouds |
| topic | deep learning semantic segmentation semantic completion indoor 3D reconstruction |
| url | https://www.mdpi.com/1424-8220/20/23/6916 |
| work_keys_str_mv | AT ingecoudron semanticextractionofpermanentstructuresforthereconstructionofbuildinginteriorsfrompointclouds AT stevenputtemans semanticextractionofpermanentstructuresforthereconstructionofbuildinginteriorsfrompointclouds AT toongoedeme semanticextractionofpermanentstructuresforthereconstructionofbuildinginteriorsfrompointclouds AT patrickvandewalle semanticextractionofpermanentstructuresforthereconstructionofbuildinginteriorsfrompointclouds |