无人机的人脸检测与跟踪系统  

Face detection and tracking system on UAV

作　　者：叶文煜 贾茜[1] YE Wenyu;JIA Qian(College of Artificial Intelligence,Jianghan University,Wuhan 430056,China)

机构地区：[1]江汉大学人工智能学院,湖北武汉430056

出　　处：《现代电子技术》2022年第20期91-96,共6页Modern Electronics Technique

摘　　要：目标跟踪与无人机航控相结合是目前的研究热点。文中利用四旋翼飞行器搭载视觉硬件模块,设计一套人脸检测与跟踪系统。该系统飞控主板采用TM4C123GH6PM,机器视觉与图像处理部分采用OpenMV,两板间通过UART方式双向通信,由视觉系统引导飞控系统完成跟踪过程。系统首先通过人脸检测算法确定单目标的相关特征;然后,不断捕捉与之匹配的单目标的中心坐标,并根据目标位置信息与视野中心的偏移量以及无人机当前飞行高度,将坐标信息转化为无人机实际飞控参数;最后,由OpenMV发送命令控制无人机改变飞行角度和位移,从而将视野中心对准目标,以实现实时的跟踪目标,达到人脸跟踪的效果。以视觉系统主导为创新,在实体无人机上的运行测试结果表明,所设计系统的跟踪达到了前期预测的标准,可以实时地完成跟踪任务。The combination of target tracking and UAV flight control is a current research hotspot. A set of face detection and tracking system is designed by means of vision hardware module mounted on quad-rotor aircraft. The TM4C123GH6PM is used as flight control main board,and the OpenMV is used for machine vision and image processing part. The two-way communication between two boards is conducted in UART mode,and the vision system is used to guide the flight control system to complete the tracking process. In the system,the face detection algorithm is used to determine the relevant features of a single target,the central coordinate of the single target matching with the features is hunted continuously,and the coordinate information is converted into the actual flight control parameters of the UAV according to the offset between the target position information and the center of the view field as well as the current flight height of the UAV. The OpenMV is used to send commands to control the UAV to change its flight angle and displacement,so as to align the center of the view field to the target to achieve real-time tracking of the target and the effect of face tracking. By taking the leading of the vision system as the innovation,the running test results on the physical UAV show that the tracking of the designed system has reached the standard of early prediction and can complete the tracking task in real time.

关 键 词：无人机 人脸检测 目标跟踪 计算机视觉 双向通信 相关特征确定 人脸跟踪 飞行控制 

分 类 号：TN820.4-34[电子电信—信息与通信工程] TP29[自动化与计算机技术—检测技术与自动化装置]