Ortho-rectified Facade Image by Fusion of 3d Laser Data and Optical Images

Obtaining three-dimensional models of urban scenes from sensor data is a challenging topic. The construction of 3D city models requires fusion of data coming from different sensors, such as laser scanners and cameras. At a first phase, we will process building facades, which will be considered as textured 3D planes. The acquisition system is equipped with two laser scanners and many (twelve) cameras. The laser scanner provides 3D point clouds of a huge size. We detail how to segment the point cloud in order to compress the information to be included in the final model and to extract the principal plane. Afterwards, we compute the homography that maps the camera image to the facade image (image fixed on the facade porter plane). The important point in our approach is that the obtained facade images are totally geo-referenced, i.e. the 3D position of each pixel is known in a global (world) reference frame.