中国空间科学技术 ›› 2021, Vol. 41 ›› Issue (4): 127-133.doi: 10.16708/j.cnki.1000-758X.2021.0059

• 论文 • 上一篇    下一篇

基于卫星遥感的岛礁影像多类目标智能化提取

江志军,张舒豪,肖进胜   

  1. 1 钱学森空间技术实验室,北京100094
    2 中国航天科技集团有限公司航天系统发展研究中心,北京100094
    3 武汉大学电子信息学院,武汉430072
  • 出版日期:2021-08-25 发布日期:2021-07-30
  • 基金资助:
    国家自然科学基金(61471272); 中国航天科技集团有限公司科技创新研发项目

Intelligent extraction of multiple objects in island and reef images based on satellite remote sensing

JIANG Zhijun,ZHANG Shuhao,XIAO Jinsheng   

  1. 1 Qian Xuesen Laboratory of Space Technology, Beijing 100094, China
    2 Aerospace System Development Research Center, China Aerospace Science and Technology Corporation, Beijing 
    100094, China
    3 School of Electronic Information, Wuhan University, Wuhan 430072, China
  • Online:2021-08-25 Published:2021-07-30

摘要: 卫星遥感影像具有背景复杂、目标尺度不一、观测方向各异、纹理不清晰等特点,主流的深度学习目标检测算法不能直接适用于卫星遥感影像的目标检测。改进了RetinaNet,使其适用于卫星遥感影像。首先设计了一种新的特征融合方式,融合ResNet50输出的特征图,使得融合后的特征图同时具有高层语义信息和低层纹理细节信息。为了减弱遥感影像复杂背景对目标特征的影响,设计了特征感知模块,在减弱噪声对特征图影响的同时增强有用特征。挑选DOTA数据集中船只、飞机和存储罐图像进行训练和测试。改进的算法与RetinaNet相比,飞机、船只和存储罐的平均精度分别提高了41%、25%、24%。基于高分二号卫星(GF2)真实影像数据的试验结果表明,提出的算法能够用于卫星遥感岛礁影像的多类目标智能化提取。

关键词: 深度学习, 遥感影像, 目标检测, 特征融合, 特征感知模块, DOTA数据集

Abstract: Satellite remote sensing images have the characteristics of complex background, different object scales, different directions, and unclear texture. The mainstream object detection algorithms based on deep learning cannot be directly applied to target detection in satellite remote sensing images. RetinaNet was improved to make it suitable for satellite remote sensing images. First, a new feature fusion method was designed to fuse the feature maps output by ResNet50, so that the fused feature maps have both highlevel semantic information and lowlevel texture detail information. Then, in order to reduce the influence of the complex background of the remote sensing image on the object features, a feature perception module was designed to reduce the influence of noise on the feature map while enhancing the useful features. Images of ship, plane and storagetank in the DOTA dataset were selected for training and testing. Experimental results show that, compared with RetinaNet, the improved algorithm has increased the average accuracy of plane, ship and storagetank by 4.1%, 2.5% and 2.4% respectively. Experimental results based on real image data of GF-2 satellite show that the proposed algorithm can be used to intelligently extract multiple types of targets in island and reef images from satellite remote sensing.

Key words: deep learning, remote sensing imagery, object detection, feature fusion, feature perception module, DOTA dataset