Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
246303 | Automation in Construction | 2015 | 10 Pages |
•Automatic surface detection from Mobile Laser Scanning point clouds•The algorithm is able to detect planar and quasi-planar surfaces.•Point clouds are simplified and structured into line clouds.•On average, 90% of the points from each surface are detected.•More than 99% of the points are correctly assigned to their surface.
An algorithm for automatic detection of planar and quasi-planar surfaces from Mobile Laser Scanning (MLS) data is proposed. The method uses line clouds for efficient data reduction of point clouds from MLS. The singular geometry of the MLS data on planar surfaces is used to transform the original point cloud into a more structured line cloud, which allows the simplification of the initial data and identification of surfaces by grouping lines. From each profile in the original dataset, strings of aligned points are identified, and a line cloud is defined by the endpoints of these strings. Lines are subsequently grouped following a set of parallelism, proximity and merging rules. The algorithm was tested using an urban dataset, and validated on 27 surfaces, by assessing the correctness and completeness of the point and line grouping. Correctness was, in all the surfaces, higher than 99%, and completeness was 90% on average.