| Automatic Recognition of Floor Plans Segments from GeoSLAM Point Cloud Data |
| Paper ID : 1246-SMPR |
| Authors: |
|
marziyeh pouraghdam1, Mohammad Sadat seresht *2 1study Surveying and Geospatial Engineering, University of Tehran,Tehran, Iran 2school Surveying and Geospatial Engineering,University of Tehran, Tehran, Iran |
| Abstract: |
| In recent years, the application of interior and exterior model of buildings has been become more and more widely in the field of Surveying and Mapping. Therefore, various approaches have been made in this area. In this regard, one of the most important application of this information is the production of the 2D floor plan and legal map. This paper presents a method for extracting a 2D floor plan from interior GeoSLAM 3D point cloud. The general process of the proposed algorithm is as follows: Initially, we select a part of the data for a room to examine the proposed algorithm. In the next step, the voxel generation was performed. Then the optimal section of the voxel cube was identified and finally, the linear structures and walls were identified and extracted using the RANSAC algorithm. Our experiments on the data collection of the University of Tehran building achieve promising results. In this study for the voxel generation step, the voxel size of 0.1 meters was determined as the optimal value. Although segments less than 30 cm could not be identified, the main linear segments of the room are extracted successfully. |
| Keywords: |
| GeoSLAM, Point Cloud, Segmentation, Floor Plan, Indoor Recognition, RANSAC |
| Status : Conditional Accept (Oral Presentation) |
