欢迎访问中国科学院大学学报,今天是

中国科学院大学学报 ›› 2022, Vol. 39 ›› Issue (2): 217-223.DOI: 10.7523/j.ucas.2020.0002

• 电子信息与计算机科学 • 上一篇    

基于核相关滤波视觉检测的多旋翼无人机对地目标跟踪与逼近

王贺1,2, 卜智勇1,3, 谭冲1   

  1. 1 中国科学院上海微系统与信息技术研究所, 上海 200050;
    2 中国科学院大学, 北京 100049;
    3 中国科学院无线传感网与通信重点实验室, 上海 200335
  • 收稿日期:2020-01-06 修回日期:2020-04-08 发布日期:2021-05-31
  • 通讯作者: 王贺
  • 基金资助:
    中国科学院青年创新促进会(2018269)资助

Ground target tracking and approaching based on multi-rotor drone and kernel correlation filtered visual detection

WANG He1,2, BU Zhiyong1,3, TAN Chong1   

  1. 1 Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, Shanghai 200050, China;
    2 University of Chinese Academy of Sciences, Beijing 100049, China;
    3 Key Laboratory of Wireless Sensor Networks and Communications, Chinese Academy of Sciences, Shanghai 200335, China
  • Received:2020-01-06 Revised:2020-04-08 Published:2021-05-31

摘要: 当前,无人机的对地目标跟踪和逼近是军民领域的研究热点之一。针对基于无人机平台对地目标跟踪与逼近中的相对运动、导引控制等问题,选用大疆多旋翼飞行器平台,设计基于Android的地面站系统。该系统使用改进的核相关滤波视觉检测算法完成对地目标跟踪与位置锁定,利用无人机与目标的距离作为尺度因子,实现多尺度跟踪,融合FHOG+Lab+LBP特征表征目标以增加跟踪器的鲁棒性。采用比例导引律控制多旋翼飞行器,完成对地目标跟踪与逼近任务。实验表明,改进的算法相比原算法在成功率和精确度上均有提高,同时验证比例导引能够控制多旋翼无人机完成跟踪与逼近任务。

关键词: 目标跟踪, 比例导引, 相关滤波

Abstract: Currently, ground target tracking and approaching based on drone is one of the hot spots for research in the military and civilian sectors. Aiming at the problems of relative motion and guidance control in the tracking and approaching of ground targets based on UAV platform, DJI multi-rotor aircraft platform was selected to design a ground station system based on Android. The system uses an improved kernel correlation filtering visual detection algorithm to complete ground target tracking and position locking. It uses the distance between the drone and the target as a scale factor to achieve multiscale tracking, and integrates FHOG + Lab + LBP features to represent the target to increase the tracker.Robustness. Proportional guidance law is used to control the multi-rotor aircraft to complete the target tracking and approaching tasks. The experimental surface improved algorithm has improved success rate and accuracy compared with the original algorithm. At the same time, the use of proportional guidance can control the multi-rotor drone to complete the tracking and approaching tasks.

Key words: target tracking, proportional guidance, correlation filtering

中图分类号: