An Accurate and Stable Pose Estimation Method Based on Geometry for Port Hoisting Machinery

Abstract

As a visual measurement method, photogrammetry is widely used in engineering measurement fields. Photogrammetry must first complete camera pose estimation. However, due to limitations of the port condition, photographs often have problems, such as attitude angles that are too large, coplanar or noncoplanar control point processing, and fewer control points, which leads to the fact that the existing algorithm may not be able to satisfy high-accuracy pose estimation. To solve this problem, a new pose estimation method is proposed in this paper. First, estimate the first pose by iterating between the optimal position and the attitude of the camera. Second, use geometric relationships and physical meaning to obtain the closest value to the actual value of the camera's attitude. Third, use the approximate camera attitude as the initial value of the iterative algorithm to obtain an accurate camera pose. Finally, the proposed method is compared with the best currently employed optimization methods through simulated and real crane image experiments. The experimental results show that our method works accurately, stable, and with no severe pose jump occurrences and can be used for pose estimation of port hoisting machinery.