Voxel-based vegetation change detection using multi-source data

Abstract. Detecting vegetation changes finds its application in several important areas, including city planning and urban science, climate change and ecological research. Several sensors and approaches can be used to measure the 3D geometry of vegetation, including aerial, mobile and terrestrial laser scanning, and photogrammetry. Other historical data sources, such as 2D shapefiles might also be available, however, the use of multi-source data presents challenges for vegetation change detection. This study presents a voxel-based approach to vegetation change detection from multi-source datasets, including laser scanning and 2D shape files from different years. A novel Octree data structure is utilised in this work that supports different operations for efficient vegetation change. We demonstrate the strengths of the approach with a case study to discuss the challenges and the future directions.