计算机应用 ›› 2020, Vol. 40 ›› Issue (5): 1529-1533.DOI: 10.11772/j.issn.1001-9081.2019091694

• 应用前沿、交叉与综合 • 上一篇    下一篇

基于深度卷积神经网络的舰载机目标检测

朱兴动1, 田少兵2, 黄葵2, 范加利2, 王正2, 陈化成2   

  1. 1.海军航空大学 岸防兵学院,山东烟台 264001
    2.海军航空大学(青岛校区) 舰面航空保障与场站管理系,山东青岛 266041
  • 收稿日期:2019-10-09 修回日期:2019-12-13 出版日期:2020-05-10 发布日期:2020-05-15
  • 通讯作者: 田少兵(1995—)
  • 作者简介:朱兴动(1967—),男,海南文昌人,教授,博士,主要研究方向:装备保障信息化; 田少兵(1995—),男,陕西宝鸡人,硕士研究生,主要研究方向:机器视觉、深度学习; 黄葵(1967—),女,浙江兰溪人,副教授,硕士,主要研究方向:装备保障信息化; 范加利(1984—),男,江苏宿迁人,讲师,博士,主要研究方向:舰面航空保障; 王正(1970—),男,山东胶州人,副教授,硕士,主要研究方向:装备保障信息化; 陈化成(1988—),男,黑龙江哈尔滨人,主要研究方向:舰载机装备保障。

Target detection of carrier-based aircraft based on deep convolutional neural network

ZHU Xingdong1, TIAN Shaobing2, HUANG Kui2, FAN Jiali2, WANG Zheng2, CHENG Huacheng2   

  1. 1.Coast Guard Academy, Naval Aviation University, YantaiShandong 264001, China
    2.Department of Ship Surface Aviation Support and Station Management, Naval Aviation University (Qingdao Campus),QingdaoShandong 266000, China
  • Received:2019-10-09 Revised:2019-12-13 Online:2020-05-10 Published:2020-05-15
  • Contact: TIAN Shaobing, born in 1995, M. S. candidate. His research interests include machine vision,deep learning.
  • About author:ZHU Xingdong, born in 1967, Ph. D., professor. His research interests include equipment support informationization.TIAN Shaobing, born in 1995, M. S. candidate. His research interests include machine vision,deep learning.HUANG Kui, born in 1967, M. S., associate professor. Her research interests include equipment support informationization.FAN Jiali, born in 1984, Ph. D., lecturer. His research interests include ship surface aviation support.WANG Zheng, born in 1970, M. S., associate professor. His research interests include equipment support informationization.CHEN Huacheng, born in 1988. His research interests include carrier-based aircraft equipment support.

摘要:

针对航母甲板面舰载机密集易遮挡,舰载机目标难以检测,且检测效果易受光照条件和目标尺度影响的问题,提出了一种改进的更快的区域卷积神经网络(Faster R-CNN)舰载机目标检测方法。该方法设计了带排斥损失策略的损失函数,并结合多尺度训练,利用实验室条件下采集的图片对深度卷积神经网络进行训练并测试。测试实验显示,相对于原始Faster R-CNN检测模型,改进后的模型对遮挡舰载机目标具有良好的检测效果,召回率提高了7个百分点,精确率提高了6个百分点。实验结果表明,所提的改进方法能够自动全面地提取舰载机目标特征,解决了遮挡舰载机目标的检测问题,检测精度和速度均能够满足实际需要,且在不同的光照条件和目标尺度下适应性强,鲁棒性较高。

关键词: 舰载机目标检测, 排斥损失策略, 更快的区域卷积神经网络, 多尺度训练

Abstract:

The carrier-based aircrafts on the carrier deck are dense and occluded, so that the carrier-based aircraft targets are difficult to detect, and the detection effect is easily affected by the lighting condition and target size. Therefore, an improved Faster R-CNN (Faster Region with Convolutional Neural Network) carrier-based aircraft target detection method was proposed. In this method, a loss function with a repulsion loss strategy was designed, and combined with multi-scale training, pictures collected under laboratory condition were used to train and test the deep convolutional neural network. Test experiments show that compared with the original Faster R-CNN detection model, the improved model has a better detection effect on occluded aircraft targets, the recall increased by 7 percentage points, and the precision increased by 6 percentage points. The experimental results show that the proposed improved method can automatically and comprehensively extract the characteristics of carrier-based aircraft targets, solve the detection problem of occluded carrier-based aircraft targets, has the detection accuracy and speed which can meet the actual needs, and has strong adaptability and high robustness under different lighting conditions and target sizes.

Key words: carrier-based aircraft target detection, repulsion loss strategy, Faster Region with Convolutional Neural Network (Faster R-CNN), multi-scale training

中图分类号: