Reconstruction 3d-models of Old Beijing City by Stuctured Light Scanning

The aim of this paper is to present a method that can be used for getting better point cloud by structured light photogrammetry. The data is from the project of 3D reconstruction of old Beijing model. The object we scanned is the 1:1000 model of old Beijing city of 1950’s. The main contents of the model are a large number of building and house models. The distinguishing feature of structured light photogrammetry is that the edge part in the models such as the edge of housetop, the intersection wall is all missed in the point cloud. Structured light photogrammetry has acquired the stereo image pair of the object. The resolution of the images is well enough to get the edges and the information can be extracted from the image. In our research, first, edge extraction method is used to extract the edge in each two image of a stereo image pair respectively. Second, based on epipolar constraint, the corresponding image pixel are determined only from the edge extraction result. The 3D coordinates of each corresponding image pixel point pair are calculated using the functions of photogrammetry and the edge points are acquired. The method is used in the old Beijing reconstruction project. Little amount data show the method is Effective. But large dataset data need to be tested.