Automatic Building Boundary Extraction from Airborne LiDAR Data

• Autor(a) principal: Carrilho, André Caceres [UNESP]
• Data de Publicação: 2019
• Outros Autores: dos Santos, Renato César [UNESP], Pessoa, Guilherme Gomes [UNESP], Galo, Mauricio [UNESP]
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.14393/rbcv71n3-46515; http://hdl.handle.net/11449/247585
• topic: Airborne LiDAR data; Building boundary extraction; α-shape

This paper aims to evaluate the quality of the automatic extraction of building contours from airborne LiDAR data in order to verify if the errors obtained in the building roof areas are compatible with the criteria established by the Brazilian Law n° 10.406, January 10th, 2002. Two approaches were compared: one present in LAStools software (lasboundary) and the other being a variant of the α-shape algorithm, whose parameter α is estimated adaptively for each building. In addition, both approaches were applied to two datasets, one with an average point density of 5.8 pts/m2 and the other with 12.5 pts/m2. In this work three quality parameters were computed: F-score, relative error in area, and PoLiS metric. The assessment of the results indicates that most of the results met the criteria established by the above-mentioned law. The variant of α-shape algorithm seems to be better than the lasboundary, indicating that the dynamic approach to determine the parameter α contributes to the quality, especially on complex buildings. Finally, the use of the point cloud with higher average point density results in higher accuracy in the building boundary extraction.