BUILDING CHANGE DETECTION USING LIDAR AND IMAGERY DATA

摘要

This paper combines LIDAR point clouds and images for change detection. The LIDAR point clouds provide the 3D information to find the changed areas. The images are employed to screen out the vegetation areas to avoid the wrong recognition in the building change detection process. The airborne imagery data includes large format camera (LFC) images and medium format camera (MFC) images. This study compares the results of vegetation areas exclusion using those two image types. For the LFC images with NIR band information, the vegetation areas are detected by the Normalized Difference Vegetation Index (NDVI). On the other hand, the MFC image does not have the NIR band. Therefore, the greenness index (GI) is employed for vegetation areas exclusion. In addition, we also use the echo ratio (ER%) of the LIDAR point clouds to exclude vegetation areas. In the same test area, we compare the building change parts by (1) LIDAR and LFC images, (2) LIDAR and MFC images, and (3) LIDAR and the ER% information. The Experimental results indicate that the LMC images reached the best results. It also shows that the MFC images and the ER% of the LIDAR point clouds could provide enough information for vegetation areas exclusion.