Research Of Relative Orientation Based On Plücker Lines For Remote Sensing Images

As a basis for stereo photogrammetry, relative orientation plays an important role in 3D reconstruction techniques, 3D visualization of digital video, high-precision positioning, and the like.For the core issue of how to realize relative orientation, the traditional method based on Euler angles using a series of well-defined corresponding points is currently used more often. There are some certain kind of limitations in the traditional method. On one hand, the rotation and translation matrices are separately solved to describe the coordinate transformation. However, the position and orientation of the captured images changes along with the spatial motion of the remote sensing platform, the separately treating may lead to the integrity of coordinate movement be violated. On the other hand, in the field of low-altitude photogrammetry, architectural photography measurement, the absence of obvious presence of the image feature points often bring problems in extracting corresponding points, and the traditional method is difficult to apply. In addition, small-light remote sensing platforms have been widely used as a strong complement to aerial approaches in recent years,and digital close-range photogrammetry as a new developing field of photogrammetry has also attracted more attention. The geometric processing of images with large attitude angles is currently facing higher demand. How to solve the above problems is now a focus in both photogrammetry and remote sensing image geometry processing, and it's also the main research direction of this article.This paper presents a new theory and method for relative orientation based on dual vector as math tool. On the basis of in-depth study of the expression of space lines and its mathematical description of topological relations, the Plücker relative orientation models based on imaging rays and joint features of line and point have been established. The corresponding light rays and corresponding feature lines are denoted by Plücker coordinates, and with dual quaternion to describing the space screw displacement, the fundamental equation of the coplanarity condition is presented, which can ultimately achieve an integrated solution. The experimental results show that the proposed methods are reliable with less dependence on initial values, and applicable to images in both small and large attitude angles with fewer iterations and a higher convergence rate. This paper can provides a new approach to the construction of geometric models in photogrammetry, which may benefit photogrammetric theory and methodology.