A Method for Determining Orientation Parameters in Image-based Reconstruction of Architectural Scenes

This paper describes a method for the determination of orientation parameters of the uncalibrated images involved in the image-based reconstruction of architectural scenes.This approach is implemented in a collection of the geometrically intuitive vanishing point method and conditional least squares adjustment with unknown parameters algorithm.We exploit the abundant parallel and orthogonal object lines to find out the vanishing point for the main direction and sequentially lead to the initial orientation parameters values.Then,more precise and stable values are found by an optimizing algorithm,based on conditional least squares adjustment with unknown parameters.During the iterative process in this algorithm,orientation parameters of the image are recalculated.Last,these orientation parameters are used to reconstruct textures landscape of the buildings by combining the vector and texture data in image.