Tracking Features with Global Motion Compensation for Drone Camera Servoing

This paper deals with visual servoing for a pan and tilt camera embedded in a drone. Video is transmitted to the ground where images are processed on a PC, and turret controls are sent back to the drone. The objective is to track any fixed object on the ground without knowledge about shape or texture and to keep it centered in the image. In order to achieve this task an algorithm that combines feature-based and global motion estimation is proposed. This algorithm provides a good robustness to very strong video transmission noise and works at a frame rate close to 25 fps. The control of the system is based on a double closed loop, which achieves a fast convergence to the desired position. Experimentation in real conditions shows the effectiveness of the proposed scheme.