Digital Building Model Generation through the Intergration of Aerial LiDAR and Multiple Photo Data in Urban Areas

Reliable and accurate Digital Building Model (DBM) generation is essential for many applications that involve in using of digital 3-D city models. This paper focuses on the integration of LiDAR and multiple aerial image data for accurate generation of DBM in urban areas. First, the LiDAR data are segmented to derive rough building boundaries, which are projected onto the multiple aerial photos to building a buffer area for each building in the proceeding procedures. Second, straight line segments are detected in the buffer area, respectively Then, straight line matching in 3D planar patch are conducted on multiple images so as to derive 3d matched lines in the planar patch to compete for the correct boundary as many as possible. Several constraints are taken into consideration in the process of matching. Finally, a similarity measure is defined to select one straight line segment which is believed to represent the building boundary. Afterwards, a closed polygon is generated for each concerned building. The proposed methodology is implemented using real datasets, and the experimental results show that the integration of LiDAR and multiple aerial photos is promising for digital building model generation.