研究期間：10108~10207;In recent years, 3D city modeling is one of the important topics in geoinformatics research. The most important reconstruction target in the research is building models. In the city scale, the 3D building models reconstruction needs plenty of time, human resources and cost. Considering the changes over a period of time, it also needs to reconstruct the buildings only for the change regions. Therefore, the main goal of this investigation is to propose an effective building modeling procedure by the fusion of aerial imagery and Lidar data to update the original building models. This project will perform for three years. In the first year, the multi-source data registration is the main objective. The data sets include aerial imagery, Lidar data and 3D building models. Because of the different acquisition time, it may exist geometric discrepancy among those data sets. We, thus, propose an automatic registration procedure. It includes two parts, one is to analyze the features in parameter space and the other is the geometric distance analysis. The reconstruction process then follows the registration. It has two stages, the first one is the change detection followed by the 3D building reconstruction. The task of the second year is to detect changed building regions using the latest aerial imagery and Lidar data. In the third year, we will integrate aerial imagery and Lidar data for 3D building reconstruction in changed regions. The step will reconstruct the details on the roof top of building models.