基于深度学习和相关滤波的飞机跟踪方法 |
投稿时间:2021-08-10 修订日期:2021-08-20 点此下载全文 |
引用本文:曹文武,陈小林,朱鑫鑫,王博,吴志佳,王雨青.基于深度学习和相关滤波的飞机跟踪方法[J].红外,2021,42(12):34~44 |
摘要点击次数: 742 |
全文下载次数: 442 |
|
|
中文摘要:针对飞行员的飞行训练任务进行了飞机的稳定跟踪。为了解决远距离下飞机目标小、天气环境复杂、飞鸟干扰等问题,提出了一种基于深度学习和相关滤波的飞机跟踪测量方法。首先选取骨干网络并建立深度学习的算法模型,接着利用大量飞机图像得到用于实际场景的参考模型,再将模型检测到的特征与相关滤波结合,从而达到飞机的稳定跟踪效果并生成目标脱靶量。然后根据跟踪和脱靶量信息,开启激光器并利用激光测距原理来测量飞机的实时距离。最后进行基于光电经纬仪的飞机捕获与跟踪实验,以验证模型和算法的有效性和可行性。实验结果表明,通过深度学习和相关滤波获得的目标信息可以用于远距离飞机的捕获与跟踪,成功地消除了复杂环境和飞鸟的干扰,实现了飞机的稳定跟踪。 |
中文关键词:图像跟踪 深度学习 激光测量 飞行训练任务 |
|
Airplane Tracking Method Based on Deep Learning and Correlation Filtering |
|
|
Abstract:The aircraft stability tracking is carried out for the pilot′s training mission.In order to solve the problems such as small target, complex weather environment and bird disturbance, an aircraft tracking measurement method based on deep learning and correlation filtering is proposed in this paper. Firstly, the backbone network is selected and the algorithm model of deep learning is established. Then, a reference model for the actual scene is obtained by using a large number of aircraft images. And the features detected by the model are combined with the correlation filtering to achieve the stable tracking effect of the aircraft and generate the missing distance of the target.According to the tracking information and missing distance, the laser is turned on and the laser ranging principle is used to measure the real-time range of the aircraft. Finally, an aircraft capture and tracking experiment based on the photoelectric theodolite is carried out to verify the validity and feasibility of the model and algorithm.The experimental results show that the target information obtained by deep learning and correlation filtering can be used to capture and track long-range aircrafts. It successfully eliminates the interference of complex environment and birds, and the stable tracking of aircraft realizes. |
keywords:image tracking deep learning laser measurement flight training task |
查看全文 HTML> 查看/发表评论 下载PDF阅读器 |