Towards an unsupervised large-scale 2D and 3D building mapping with airborne LiDAR data

A 2D and 3D building map provides invaluable information for understanding human activities and their impacts on the Earth and its environment. Despite enormous efforts to improve the quality of building maps, current large-scale building maps generated from automated methods have many errors and uncertainties and are often limited to providing only 2D building information. This study presents an open-source unsupervised 2D and 3D building extraction algorithm with airborne LiDAR data that is suitable for large-scale building mapping. Our algorithm operates in a fully unsupervised manner and does not require either any training label or training procedure. Our algorithm consists of simple operations of morphological filtering and planarity-based filtering. Thus, the computation is efficient, and the results are easy to predict, which can greatly reduce uncertainties in the resulting building map. A quantitative and qualitative evaluation with a large-scale dataset (> 550 $km^2$) of Denver and New York City showed that our algorithm can produce more accurate building maps than Microsoft Building Footprints which is generated by a deep learning-based method. Extensive evaluations in different conditions of landscapes confirmed that our algorithm is scalable and can be improved further with appropriate parameter selection. We also detailed the impact of parameters and potential sources of error to assist potential users of our algorithm. Our LiDAR-based algorithm has advantages in that it is computationally efficient to generate both 2D and 3D building maps, while it generates accurate and explainable results. Our proposed algorithm provides great potential towards a global-scale 2D and 3D building mapping with airborne LiDAR data.