Contours-Matching in Building Reconstruction from a Single Image

Image-based building reconstruction is a hot point in computer vision and computer graphics, but few works have done on reconstruction from a single image because it is an ill-conditional problem and is very difficult to resolve. In this paper, we present an efficient method by matching contours between image and projection of 3D models. Our method simplifies the reconstruction process and avoids cameras calibrating and human interaction. Given a building photo captured in daily life as input, we segment the photo into different regions firstly, then extract the contour of the building automatically. Finally, we get 3D building model by matching the contour of the building with projection contour of 3D models from different views stored in model database. Experiments prove that our method is efficient and effective especially for building photos containing shelters.