The use of Otsu algorithm and multi-temporal airborne LiDAR data to detect building changes in urban space

Building change detection techniques are essential for several urban applications. In this context, multi-temporal airborne LiDAR data has been considered an effective alternative since it has some advantages over conventional photogrammetry. Despite several works in the literature, the automatic class definition with great accuracy and performance remains a challenge in change detection. The developed strategies usually explore training samples or empirical thresholds to discriminate the classes. To overcome this limitation, we proposed an automatic building change detection method based on Otsu algorithm and median planarity attribute computed from eigenvalues. The main contribution corresponds to the automatic and unsupervised identification of building changes. The experiments were conducted using airborne LiDAR data from two epochs: 2012 and 2014. From qualitative and quantitative analysis, the robustness of the proposed method in detecting building changes in urban areas was evaluated, presenting completeness and correctness around 99% and 76%, respectively.