Registration Of Aerial Images With Airborne LiDAR Based On Plücker Line

LiDAR technology has so many advantages include high precision, strong independence, high automation and high efficiency, so it is widely applied in many fields, for example, navigation technology, aerial photography and geological disaster prevention and prediction. Registration of aerial images with airborne LiDAR point clouds is the key to feature extraction. The existing registration methods generally transform LiDAR point clouds into 2D images, then register with aerial images, but this will bring the interpolation error. The other kind of methods match many aerial images to get 3D point clouds, then register with LiDAR point clouds, but it fails without a good initial value. Directly match the conjugated lines of LiDAR point clouds and aerial images can avoid the error mentioned above. Line feature is a high level geometric feature, which is more easily to be extracted than the point feature, and has stronger geometric topology and constraint.The existing registration methods of LiDAR point clouds with aerial images based on line features provide a weak constraint because the geometric topology of lines is not taken into full account. A registration method based on Plücker line is proposed to strengthen the geometric constraints of lines. First, conjugate lines of LiDAR point clouds and aerial images are represented in Plücker coordinates. Then determine the relative position and attitude relationship between the conjugate lines by the Plücker coplanarity equation. Last, establish the registration model of LiDAR and single image based on Plücker coplanarity condition. So the coordinate transformation of image point and the corresponding LiDAR point is achieved through the spiral motion of Plücker lines in 3D space. Based upon the above, make further research into unified position and attitude resolution of the stereo images, and construct the registration model of LiDAR and stereo images based on Plücker line to improve the geometric positioning accuracy. The experimental results show that the proposed model solves the position and attitude parameters simultaneously with high precision. The proposed model is compact and clear with obvious geometric meaning and strong geometric constraint, and jointly describes the rotation and translation to avoid coupling error and approve the accuracy. This research provides technical support for high-quality 3D earth spatial information acquisition.