基于双变焦摄像头的空中目标跟拍系统

Air Target Tracking System Based on Two Zoom Cameras

针对空中目标的跟拍画面不清晰、不稳定的问题,设计了一种基于双变焦摄像头的空中目标跟拍系统,采用高精确云台、双变焦摄像头及高性能计算机,能够根据目标距离动态调整双变焦摄像头的焦距,在跟踪目标并拍摄清晰、稳定的运动图像。该系统采用了基于深度学习的目标检测算法,通过加权质心算法计算目标质心位置;采用卡尔曼滤波与轨迹拟合相结合的位置预测算法,利用三面交汇原理计算目标的空间位置。实验结果表明:该系统能够获取目标稳定清晰的图像并计算目标的空间位置,验证了系统的可靠性和实用性。

Aimed at the problem of indistinct and unstable tracking picture of air target,an air target tracking system based on two zoom cameras is designed.The system,equipped with a high-precision pan tilt,two zoom cameras and a high-performance computer,can dynamically adjust the focal length of the dual zoom camera in line with the target distance,and capture a clear and stable moving image of the target while tracking the target.The system adopts the target detection algorithm based on deep learning and calculates the target centroid position through the weighted centroid algorithm.The system also employs the position prediction algorithm combining Kalman filter and trajectory fitting to calculate the spatial position of the target by three-sided intersection principle.The experimental results show that the system can obtain a stable and clear image of the target and accurately calculate the spatial position of the target,which verifies the reliability and practicability of the system.