无人机自主目标识别与定位应用研究

Research on the application of UAV autonomous target recognition and positioning

摘要:

本文面向无人机自主侦察任务中在线目标识别与定位需求,首先梳理了无人机侦察中目标识别领域的相关研究成果。然后,介绍了Faster RCNN目标识别算法的实现原理,并针对任务需求进行了改进。之后,介绍了图像拼接的相关算法并进一步提出了目标相对定位算法。最后,设计了完整的侦察试验流程对所设计自主目标识别与定位方法进行验证。结果表明,本文改进的目标检测网络能够达到83.3%的识别准确率和35帧/秒的识别速度,所提出的相对定位算法可以达到0.702m的平均定位精度,能够满足侦察无人机在线目标识别与定位的任务需求。

Abstract:

Aiming at the requirements of online target recognition and positioning in UAV autonomous reconnaissance mission, this paper firstly summarizes relevant research results in the field of target recognition in UAV reconnaissance. Then, the implementation principle of Faster RCNN is introduced and the improvement is made according to the task requirements. After that, we introduce the correlation algorithm of image mosaic and propose the target relative positioning algorithm. Finally, a complete outdoor test procedure is designed to verify the proposed method. The results show that the improved target detection network in this paper can achieve the recognition accuracy rate of 85.6% and the recognition speed of 35 frames per second, and the proposed relative positioning algorithm can achieve the average positioning accuracy of 0.702m, which can meet the task requirements of the online target identification and positioning of reconnaissance UAV.