航空学报 > 2024, Vol. 45 Issue (17): 530062-530062   doi: 10.7527/S1000-6893.2024.30062

基于残差混合监督网络的无人机目标阴影检测

王潇1, 刘贞报1(), 史忠科2   

  1. 1.西北工业大学 民航学院,西安 710072
    2.西北工业大学 自动化学院,西安 710129
  • 收稿日期:2024-01-02 修回日期:2024-01-19 接受日期:2024-03-15 出版日期:2024-04-02 发布日期:2024-03-29
  • 通讯作者: 刘贞报 E-mail:liuzhenbao@nwpu.edu.cn
  • 基金资助:
    国家自然科学基金(52072309);陕西省重点研发计划(2019ZDLGY14-02-01);深圳市基础研究资助项目(JCYJ20190806152203506);航空科学基金(ASFC-2018ZC53026)

Shadow detection of UAV target based on residual mixed supervision network

Xiao WANG1, Zhenbao LIU1(), Zhongke SHI2   

  1. 1.School of Civil Aviation,Northwestern Polytechnical University,Xi’an 710072,China
    2.School of Automation,Northwestern Polytechnical University,Xi’an 710129,China
  • Received:2024-01-02 Revised:2024-01-19 Accepted:2024-03-15 Online:2024-04-02 Published:2024-03-29
  • Contact: Zhenbao LIU E-mail:liuzhenbao@nwpu.edu.cn
  • Supported by:
    National Natural Science Foundation of China(52072309);Key Research and Development Program of Shaanxi

摘要:

无人机在检测和跟踪目标过程中受到目标伪装、目标遮挡、移动躲避以及假目标等因素的干扰,而无人机目标附近的阴影区域加剧了这些因素对目标检测和跟踪性能的影响,因此检测无人机目标阴影区域是无人机领域的重要研究任务之一。现有无人机目标阴影检测方法面临训练数据数量有限、数据收集标注困难以及无人机目标中存在大量尺寸较小的细碎阴影区域等问题,针对这些问题,提出一种基于残差混合监督网络的无人机目标阴影检测算法。首先针对无人机目标阴影检测任务的特点设计分辨率注意力网络,在结合底层纹理特征和高层语义特征的过程中,更准确地保留底层纹理特征。然后设计混合监督网络扩充训练数据集,结合普通阴影检测数据集和无人机目标阴影检测数据集训练教师网络,使用无人机阴影检测数据集和教师网络的参数训练学生网络。同时设计残差图像,利用教师网络检测结果和标准结果之间的残差图像扩充训练数据集,使阴影检测网络更加关注细碎阴影区域。最后,在2个公开实验数据集上和已有方法进行对比实验,在各个评价参数上取得了最多41.6%的提升效果,证明所提无人机目标阴影检测算法较好的解决了现有方法存在的问题,具有较高的准确性。

关键词: 无人机目标, 阴影检测, 混合监督网络, 残差图像, 分辨率注意力网络

Abstract:

The targets camouflage, targets occlusion, moving dodge and fake targets deteriorate the performance of UAV object detection and tracking, and the shadow regions around UAV target aggravate the negative effect of these factors. Thus, shadow detection is an important task for UAV. The shadow detection of UAV target suffering from limited training images, difficulty in labeling ground truth data and mass of tiny shadow regions. To deal with these problems, we propose a UAV target shadow detection method based on residual mixed supervision network. Firstly, we design a resolution-aware attention shadow detection network based on the character of shadow regions in UAV target. The newly designed network can maintain the lower texture feature more accurately. Then we design mixed supervision network to enlarge the number of training images. The teacher network is trained by both ordinary dataset and UAV dataset, while the student network is trained based on UAV dataset and the parameter of teacher network. Meanwhile we design residual images to further enlarge the number of training images and makes the network pay more attention to tiny shadow regions. The residual image is calculated by measuring the difference between detection results of teacher network and ground truth data. At last, the proposed method is compared with existing methods on two public UAV target shadow detection datasets. The evaluation metrics are improved by 41.6% at most. The experiment proves the effectiveness and accuracy of proposed shadow detection method on UAV target.

Key words: UAV target, shadow detection, mixed supervision network, residual images, resolution-aware attention network

中图分类号: