BUILDING CHANGE DETECTION USING MULTI-TEMPORAL AIRBORNE LIDAR DATA

Abstract. Building change detection is essential for monitoring urbanization, disaster assessment, urban planning and frequently updating the maps. 3D structure information from airborne light ranging (LiDAR) very effective detecting changes. But point cloud LiDAR(ALS) holds an enormous amount of unordered irregularly sparse information. Handling such data tricky consumes large memory processing. Most this not necessary when we are looking a particular type change. In study, propose automatic method that reduces clouds into much smaller representation without losing required The utilizes Deep Learning(DL) model U-Net segmenting buildings background. Produced segmentation maps then processed further changes results refined using morphological methods. For task, used multi-temporal LiDAR data. acquired over Stockholm in years 2017 2019. classified four types: ‘newly built’, ‘demolished’, ‘taller’ ’shorter’. detected visualized one map better interpretation.