Abstract:Due to the narrowness and passivity of the traditional fixed camera monitoring, as well as its inability to adjust the focal length values to obtain clear images of the target, crane hook following is confronted with a significant challenge. Therefore, an active following algorithm has thus been proposed for crane hooks based on PTZ camera. Firstly, a calculation can be made of the initial state angle of the camera and the focal length value in the built-in parameter K through camera calibration. By using geometric projection relationships, a dynamic imaging model can be established between the camera and the workshop coordinate system, followed by an estimation of the P, T, and Z values of the camera based on the model and the position of the crane hooks. Then, in view of the problem of high-resolution images gradually shifting from the center area of the image, a feature point automatic alignment algorithm is adopted for a calculation of the P and T increments of the camera twice. Finally, based on the estimated P, T, Z values and their increments, the camera interface function is called to adjust the attitude, so that the crane hook is always located in the center of the image, thus completing the active following task. The experimental results show that the proposed algorithm can be well applied to target tracking in a large field of view. Meanwhile, the algorithm requires a real-time collection of the world coordinates of the target point, which makes it applicable to target tracking equipped with GPS positioning equipment or distance sensors as well.