An Embedded Automatic Focusing System For Tracking Small Flying Targets

The high-speed motion capture and image acquisition of small flying targets are of great significance in the fields of bionics, aviation safety, and flight mechanism research. Aiming at the out-of-focus problem that is easy to occur in the image tracking system, this paper designs an embedded automatic focusing system for tracking small flying targets. We improve YOLOv3 network framework to give full play to the advantages of rapid detection, and use the Laplace operator to evaluate the sharpness of the image that contains the target, and find the best focus position by using the adaptive step-size hill climbing algorithm to achieve the goal of fast recognition and automatic focusing function on small flying targets. The algorithm is implemented based on the HiSilicon 3516CV500 chip. Experiments show that the identification and focusing system effectively improves the detection speed in the case of maintaining high identification accuracy, and the real-time frame rate of target detection can reach 14fps, which provides a solution to the out-of-focus problem of the small flying target tracking system.