Three-step approach to camera pose estimation

Camera pose estimation is one of the oldest and most frequent tasks in computer vision and photogrammetry. The paper presents a three-step (namely determination of camera-point distances, pose angles and line elements) approach to camera pose estimation, and derives the basic mathematic models and introduces the computational method in each step; especially puts forward the direct calculus to determine the rotation matrix. The case study shows the approach is correct and efficient no matter the pose angles are small or large.