Identification of roofs perimeter from aerial and satellite images

The computer-driven extraction of roofs perimeter is a very important step in 3D reconstruction of buildings as it allows to reduce human interactions. In this context, very high resolution optical images, from airborne surveys or from satellites, were used in this paper with the aim of extracting such information. The proposed method is based on a color image segmentation algorithm. The seed-growing approach has been considered, thanks to its capability and sensitivity in local analysis. A multi-seed method is presented, characterized by an adaptive growing mechanism and able to estimate internal parameters on the basis of the image under analysis. A further simplification to the user is given, since an automatic extraction of seed points has been introduced, that is fundamental for a complete image segmentation. Different images have been tested and experimental results prove the effectiveness of the proposed method.