留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

基于GLNet和HRNet的高分辨率遥感影像语义分割

赵紫旋 吴谨 朱磊

赵紫旋, 吴谨, 朱磊. 基于GLNet和HRNet的高分辨率遥感影像语义分割[J]. 红外技术, 2021, 43(5): 437-442.
引用本文: 赵紫旋, 吴谨, 朱磊. 基于GLNet和HRNet的高分辨率遥感影像语义分割[J]. 红外技术, 2021, 43(5): 437-442.
ZHAO Zixuan, WU Jin, ZHU Lei. High-resolution Remote Sensing Image Semantic Segmentation Based on GLNet and HRNet[J]. Infrared Technology , 2021, 43(5): 437-442.
Citation: ZHAO Zixuan, WU Jin, ZHU Lei. High-resolution Remote Sensing Image Semantic Segmentation Based on GLNet and HRNet[J]. Infrared Technology , 2021, 43(5): 437-442.

基于GLNet和HRNet的高分辨率遥感影像语义分割

基金项目: 

国家自然科学基金青年基金项目资助 61502358

国家自然科学基金青年基金项目资助 61702384

详细信息
    作者简介:

    赵紫旋(1997-),女,湖北武汉人,硕士研究生,研究方向为图像处理、深度学习。E-mail:zhaozixuan19970708@163.com

    通讯作者:

    吴谨(1967-),女,安徽芜湖人,博士,教授,研究方向为图像处理与模式识别、信号处理与多媒体通信等。E-mail:wujin@wust.edu.cn

  • 中图分类号: TP751.1

High-resolution Remote Sensing Image Semantic Segmentation Based on GLNet and HRNet

  • 摘要: 在GLNet(Global-Local Network)中,全局分支采用ResNet(Residual Network)作为主干网络,其侧边输出的特征图分辨率较低,而且表征能力不足,局部分支融合全局分支中未充分学习的特征图,造成分割准确率欠佳。针对上述问题,提出了一种基于GLNet和HRNet(High-Resolution Network)的改进网络用于高分辨率遥感影像语义分割。首先,利用HRNet取代全局分支中原有的ResNet主干,获取表征能力更强,分辨率更高的特征图。然后,采用多级损失函数对网络进行优化,使输出结果与人工标记更为相似。最后,独立训练局部分支,以消除全局分支中特征图所带来的混淆。在高分辨率遥感影像数据集上,对所提出的改进网络进行训练和测试,实验结果表明,改进网络在全局分支和局部分支上的平均绝对误差(Mean Absolute Error,MAE)分别为0.0630和0.0479,在分割准确率和平均绝对误差方面均优于GLNet。
  • 图  1  GLNet和本文改进网络的流程对比

    Figure  1.  The comparison of GLNet and network structure proposed in this paper

    图  2  HRNet的基本网络结构

    Figure  2.  Basic Network structure of HRNet

    图  3  HRNet的交换单元

    Figure  3.  Exchange unit of HRNet

    图  4  GLNet的全局分支结构

    Figure  4.  The structure of global branch

    图  5  GLNet的局部分支网络结构

    Figure  5.  The structure of local branch of GLNet

    图  6  聚合过程

    Figure  6.  The process of aggregation

    图  7  高分辨率遥感图像分割结果

    Figure  7.  The segmentation results of high resolution remote sensing images

    图  8  高分辨率遥感图像PR曲线图

    Figure  8.  The PR curves of high-resolution remote sensing image

    表  1  Global分支实验对比

    Table  1.   Comparison of global branch experiments

    MAE GPU Memory/M
    GLNet 0.0730 1980
    Ours 0.0630 2715
    下载: 导出CSV

    表  2  Local分支实验对比

    Table  2.   Comparison of local branch experiments

    MAE GPU Memory/M
    GLNet 0.0572 1900
    Ours 0.0479 1869
    下载: 导出CSV
  • [1] 田萱, 王亮, 丁琪. 基于深度学习的图像语义分割方法综述[J]. 软件学报, 2019, 30(2): 250-278. https://www.cnki.com.cn/Article/CJFDTOTAL-RJXB201902014.htm

    TIAN X, WANG L, DING Q. Review of Image Semantic Segmentation Based on Deep Learning[J]. Journal of Software, 2019, 30(2): 440-468. https://www.cnki.com.cn/Article/CJFDTOTAL-RJXB201902014.htm
    [2] LONG J, Shelhamer E, Darrell T. Fully convolutional networks for semantic segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2015: 3431-3440.
    [3] Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition[C]//Proceedings of International Conference on Learning Representation, 2015: 1-13.
    [4] Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for biomedical image segmentation[C]//Proceedings of International Conference on Medical Image Computing and Computer-assisted Intervention, 2015: 234-241.
    [5] Noh H, Hong S, Han B. Learning deconvolution network for semantic segmentation[C]//Proceedings of the IEEE International Conference on Computer Vision, 2015: 1520-1528.
    [6] Badrinarayanan V, Kendall A, Cipolla R. Segnet: A deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(12): 2481-2495. doi:  10.1109/TPAMI.2016.2644615
    [7] CHEN L C, Papandreou G, Kokkinos I, et al. Deep Lab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, 40(4): 834. doi:  10.1109/TPAMI.2017.2699184
    [8] Paszke A, Chaurasia A, Kim S, et al. Enet: A deep neural network architecture for real-time semantic segmentation[J/OL]. arXiv preprint, 2016, https://arxiv.org/abs/1606.02147.
    [9] ZHAO H, QI X, SHEN X, et al. Icnet for real-time semantic segmentation on high-resolution images[C]//Proceedings of the European Conference on Computer Vision, 2018: 405-420.
    [10] CHEN W, JIANG Z, WANG Z, et al. Collaborative global-local networks for memory-efficient segmentation of ultra-high resolution images[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2019: 8924-8933.
    [11] HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016: 770-778.
    [12] LIN T Y, Dollár P, Girshick R, et al. Feature pyramid networks for object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017: 2117-2125.
    [13] SUN K, XIAO B, LIU D, et al. Deep high-resolution representation learning for human pose estimation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2019: 5693-5703.
    [14] HOU Q, CHENG M M, HU X, et al. Deeply supervised salient object detection with short connections[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017: 3203-3212.
    [15] LIN T Y, Goyal P, Girshick R, et al. Focal loss for dense object detection[C]//Proceedings of the IEEE International Conference on Computer Vision, 2017: 2980-2988.
    [16] Kingma D P, Ba J. Adam: A method for stochastic optimization[J/OL]. arXiv preprint arXiv: 1412.6980, 2014.
    [17] ZHAO H, SHI J, QI X, et al. Pyramid scene parsing network[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017: 2881-2890.
  • 加载中
图(8) / 表(2)
计量
  • 文章访问数:  409
  • HTML全文浏览量:  112
  • PDF下载量:  64
  • 被引次数: 0
出版历程
  • 收稿日期:  2020-04-08
  • 修回日期:  2020-06-23
  • 刊出日期:  2021-05-22

目录

    /

    返回文章
    返回