An interactive registration method for images to the 3D urban scene model

In order to fast register a camera into a 3D scene model under the Manhattan-World assumption, a method of matching corresponding 2D and 3D lines based on vanishing point is proposed in this paper. Firstly, this method detects line segments and estimates three orthogonal vanishing points to determine the local length of camera and the matrix from world to camera space. Afterwards, one line is drawn in the image and 3D model respectively. The registration of the camera with respect to the model is performed by matching these two lines. To decrease the interference, we connect the fragments of lines and adopt a minimization approach to update the vanishing points. Our method makes three assumptions: 1) at least two dominant vanishing points are contained, 2) no obvious distortion, 3) the principal point lies at the center of image. At present, our approach of vanishing point estimation achieves a state-of-the-art performance on the standard test database. The entire method has been validated to be applied to augment virtual environment and fast 3D scene reconstruction.