Automatic Texture Acquisition for 3D Model Using Oblique Aerial Images

This paper describes an approach for automatic texture acquisition obtained from 3D city model and oblique aerial images. Firstly, the image feature lines should be extracted. Secondly, the extracted lines are matched with the corresponding 3D feature lines of object-space from the 3D city model, by estimating the image-space position of the 3D lines with the coarsely provided exterior orientation (EO) parameters of the image. Then, for each corresponding lines, we could list two equations based on the coplanarity constraints of object-space; that is, the accurate EO parameters can be calculated by least square method (LSM) with at least three pairs of conjugate lines. Finally, the interest image areas including the building surfaces are rectified with the refined EO parameters and the textures are obtained. In this paper, the principles of the method are described in detail along with an experiment carried out on a dataset of downtown area abroad, which has proved the correctness and robustness of such method.