Automatic Building Change Detection Using Multi-Temporal Airborne Lidar Data

The automatic detection of building changes is an essential process for urban area monitoring, urban planning, and database update. In this context, 3D information derived from multi-temporal airborne LiDAR scanning is one effective alternative. Despite several works in the literature, the separation of change areas in building and non-building remains a challenge. In this sense, it is proposed a new method for building change detection, having as the main contribution the use of height entropy concept to identify the building change areas. The experiments were performed considering multi-temporal airborne LiDAR data from 2012 and 2014, both with average density around 5 points/m2. Qualitative and quantitative analyses indicate that the proposed method is robust in building change detection, having the potential to identify small changes (larger than 20 m2). In general, the change detection method presented average completeness and correctness around 97% and 71%, respectively.