Please wait a minute...
img

官方微信

遥感技术与应用  2022, Vol. 37 Issue (4): 820-828    DOI: 10.11873/j.issn.1004-0323.2022.4.0820
深度学习专栏     
GCM+-LANet:遥感图像语义分割的全局卷积模块与局部注意力网络模型
翁梦倩1(),胡蕾1(),张永梅2,凌杰1,李云洪1
1.江西师范大学 计算机信息工程学院,江西 南昌 330022
2.北方工业大学 信息学院,北京 100144
GCM+-LANet:Global Convolution Module+ and Local Attention Network for Semantic Segmentation of Remote Sensing Images
Mengqian Weng1(),Lei Hu1(),Yongmei Zhang2,Jie Ling1,Yunhong Li1
1.Jiangxi Normal University,School of Computer Information Engineering,Nanchang 330022,China
2.North China University of Technology,School of Information,Beijing 100144,China
 全文: PDF(4128 KB)   HTML
摘要:

遥感图像地物种类丰富、尺寸多变、分布不均衡、背景复杂,导致经典图像语义分割网络难以在遥感图像上取得理想分割效果。局部注意力网络模型(LANet)在遥感图像语义分割上取得了较好的实验效果,但大尺寸、小尺寸和细长的地物目标分割效果不佳。提出了一种改进LANet网络的高分辨率遥感图像语义分割网络模型,首先,针对全局特征提取设计了全局卷积模块(GCM+),以组合卷积的形式扩大感受野,提升大尺寸地物目标的分割性能;其次,利用针对计算机视觉提出的激活函数Funnel ReLU(FReLU)来解决细小目标漏分的问题。实验结果表明:该网络模型在Potsdam数据集上平均交并比达到了75.83%,像素准确率达到了94.95%,比基础网络LANet有较大提升。

关键词: 遥感图像语义分割全局卷积模块局部注意力网络模型激活函数    
Abstract:

Because of the variety, variable size, uneven distribution and complex background of ground objects, the classical image semantic segmentation network is difficult to achieve ideal segmentation results in remote sensing images. Local Attention Network (LANet) has achieved good results in remote sensing image semantic segmentation, but the segmentation effect of large-size, slender and small ground objects are not very good. Therefore, based on LANet a semantic segmentation network is proposed to high resolution remote sensing image. Firstly, Global Convolution Module+ (GCM+) aimed at global feature extraction is designed to enlarge the receptive field by combining convolution, which can improve the segmentation performance of large-size objects. Secondly, the activation function Funnel ReLU (FReLU) proposed to computer vision is used to solve the problem of missing slender and small targets. The experimental results show that the mean intersection over union of the network of the Potsdam dataset reaches 75.83 %, and the pixel accuracy reaches 94.95 %, which is greatly improved than LANet.

Key words: Remote sensing image    Semantic segmentation    Global convolution module    Local attention network    Activation function
收稿日期: 2021-08-31 出版日期: 2022-09-28
:  TP75  
基金资助: 国家自然科学基金项目(61662033);江西省教育厅科学技术研究项目(GJJ210326)
通讯作者: 胡蕾     E-mail: 1059257750@qq.com;hulei@jxnu.edu.cn
作者简介: 翁梦倩(1996-),女,江西上饶人,硕士研究生,主要从事遥感图像语义分割研究。E?mail: 1059257750@qq.com.
服务  
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章  
翁梦倩
胡蕾
张永梅
凌杰
李云洪

引用本文:

翁梦倩,胡蕾,张永梅,凌杰,李云洪. GCM+-LANet:遥感图像语义分割的全局卷积模块与局部注意力网络模型[J]. 遥感技术与应用, 2022, 37(4): 820-828.

Mengqian Weng,Lei Hu,Yongmei Zhang,Jie Ling,Yunhong Li. GCM+-LANet:Global Convolution Module+ and Local Attention Network for Semantic Segmentation of Remote Sensing Images. Remote Sensing Technology and Application, 2022, 37(4): 820-828.

链接本文:

http://www.rsta.ac.cn/CN/10.11873/j.issn.1004-0323.2022.4.0820        http://www.rsta.ac.cn/CN/Y2022/V37/I4/820

图1  PAM模块结构图
图2  AEM模块结构图
图3  GCM+-LANet网络模型结构图
图4  GCM+结构图
图5  RestNet50的BottleNeck结构图
图6  ReLU、PReLU和FReLU激活函数(a)ReLU (b)PReLU (c)FReLU
图7  Potsdam数据集
网络模型PA/%F1/%MIoU/%
FCN92.4077.2170.09
U-Net91.8277.0569.98
SegNet91.5575.9268.62
LANet91.8377.2970.19
DeepLab V3+86.9968.1459.34
Coordinate Attention90.9475.4567.94
GCM+-LANet(k=3)94.4280.9975.27
表1  不同方法在Potsdam数据集上的分割精度
图8  不同网络模型在Potsdam数据集上的预测结果对比
图9  k值对MIoU和参数量的影响分析
1 Chao P, Zhang X, Gang Y, et al. Large kernel matters-improve semantic segmentation by global convolutional network[C]∥ Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Honolulu, USA: IEEE, 2017: 4353-4361.
2 Likas A, Vlassis N, Verbeek J. The Global K-means clustering algorithm[J]. Pattern Recognition, 2003, 36(2): 451-461.DOI:10.1016/S0031-3203(02)00060-2 .
doi: 10.1016/S0031-3203(02)00060-2
3 Carson C, Belongie S, Greenspan H, et al. Blobworld: Image segmentation using expectation-maximization and its application to image querying[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002, 24(8):1026-1038. DOI: 10.1109/TPAMI.2002.1023800 .
doi: 10.1109/TPAMI.2002.1023800
4 Li Shuang, Ding Shengyan, Qian Yuexiang. The decisiontree classification and its application research in land cover[J]. Remote Sensing Technology and Application, 2002, 17(1): 6-11.
4 李爽, 丁圣彦, 钱乐祥. 决策树分类法及其在土地覆盖分类中的应用[J].遥感技术与应用, 2002, 17(1): 6-11.
5 Song M, Civco D. Road extraction using SVM and image segmentation[J]. Photogrammetric Engineering & Remote Sensing,2004,70(12): 1365-1371. DOI: 10.14358/P-ERS. 70.12.1365 .
doi: 10.14358/P-ERS. 70.12.1365
6 Stéphane G, Olivier G. A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood[J]. Systematic Bology, 2003, 52(5): 696-704. DOI: 10.1080/10635150390235520 .
doi: 10.1080/10635150390235520
7 Jordan M I, Mitchell T M. Machine learning: Trends, perspectives, and prospects[J]. Science, 2015, 349(6245): 255-260. DOI: 10.1126/science.aaa8415 .
doi: 10.1126/science.aaa8415
8 Gu Xiaotian, Gao Xiaohong, Ma Huijuan, et al. Comparison of machine learning methods for land use/land cover classification in the complicated terrain regions[J].Remote Sensing Te-chnology and Application, 2019, 34(1):59-69.
8 谷晓天, 高小红, 马慧娟,等. 复杂地形区土地利用/土地覆被分类机器学习方法比较研究[J]. 遥感技术与应用, 2019, 34(1): 59-69.
9 Long J, Shelhamer E, Darrell T. Fully convolutional networks for semantic segmentation[C]∥ Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR),Boston,MA,USA,IEEE,2015:3431-3440.
10 Zhang H, Dana K, Shi J, et al. Context encoding for semantic segmentation[C]∥ Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition( CVPR). Salt Lake City, UT, USA, IEEE, 2018: 7151-7160.
11 Ronneberger O, Fischer P, Brox T. U-Net: convolutional networks for biomedical image segmentation[C]∥ International Conference on Medical Image Computing and Computer-Assisted Intervention. Munich, Germany: Springer,2015: 234-241.
12 Badrinarayanan V, Kendall A, Cipolla R. Segnet: A deep convolutional encoder-decoder srchitecture for image segmentation[J]. IEEETransactions on Pattern Analysisand Machine Intelligence,2017,39(12):2481-2495. DOI: 10.1109/TPAMI.2016.2644615 .
doi: 10.1109/TPAMI.2016.2644615
13 Chen L C, Papandreou G, Kokkinos I, et al. Semantic image segmentation with deep convolutional nets and fully connected CRFs[EB/OL]. arXiv Preprint arXiv: , 2014.
14 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):834-848. DOI:10.1109/TPAMI.2017.2699184 .
doi: 10.1109/TPAMI.2017.2699184
15 Chen L C, Papandreou G, Schroff F, et al. Rethinking atrous convolution for semantic image segmentation[EB/OL]. arXiv Preprint arXiv: , 2017.
16 Chen L C, Zhu Y, Papandreou G, et al. Encoder decoder with atrous separable convolution for semantic image segmentation[C]∥ Proceedings of the European Conference on Computer Vision(ECCV). Munich,Germany:IEEE,2018:801-818.
17 Yang M D, Tseng H H, Hsu Y C, et al. Semantic segmentation using deep learning with vegetation indicesfor rice lodging identification in multi date UAV visible Images[J]. Remote Sensing, 2020, 12(4): 633-652. DOI: 10.3390/rs12040633 .
doi: 10.3390/rs12040633
18 Abdollahi A, Pradhan B, Alamri A M. An ensemble architecture of deep convolutional segnet and U-Net networks for building semantic segmentation from high resolution aerial images[J].Geocarto International,2020(3):116. DOI: 10.1080/ 10106049.2020.1856199 .
doi: 10.1080/ 10106049.2020.1856199
19 He H, Yang D, Wang S, et al. Road extraction by using atrous spatial pyramid pooling integrated encoder decoder network and structural similarity loss[J]. Remote Sensing, 2019, 11(9): 1015-1030. DOI:10.3390/rs11091015 .
doi: 10.3390/rs11091015
20 Ding L, Tang H, Bruzzone L. LANet: Local attention embedding to improve the semantic segmentation of remote sensing images[J]. IEEE Transactions on Geoscience and Remote Sensing,2020,59(1):426-435. DOI:10.1109/TGRS. 2020. 2994150 .
doi: 10.1109/TGRS. 2020. 2994150
21 Hou Q, Zhou D, Feng J. Coordinate attention for efficient mobile network design[C]∥ Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR). 2021: 13713-13722.
22 Ma N, Zhang X, Sun J. Funnel activation for visual recognition[C]∥ Proceedings of the European Conference onComputer Vision(ECCV),Glasgow, UK: Springer, 2020: 351-368.
23 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(CVPR). Las Vegas, NV, USA: IEEE, 2016: 770-778.
24 Hu J, Shen L, Sun G. Squeeze and excitation networks[C]∥Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Salt Lake City, UT: IEEE, 2018: 7132-7141.
25 Rottensteiner F, Sohn G, Gerke M, et al. Results of the ISPRS benchmark on urban object detection and 3D building reconstruction[J]. ISPRS Journal of Photogrammetry and Remote Sensing,2014,93:256-271. DOI:10.1016/j.isprsjprs. 2013.10.00 .
doi: 10.1016/j.isprsjprs. 2013.10.00
[1] 付涵,范湘涛,严珍珍,杜小平. 基于深度学习的遥感图像目标检测技术研究进展[J]. 遥感技术与应用, 2022, 37(2): 290-305.
[2] 盖一铭,阿里木·赛买提,王伟,吉力力·阿不都外力. 基于样本迁移的干旱区地表覆盖快速更新[J]. 遥感技术与应用, 2022, 37(2): 333-341.
[3] 陈妮,应丰,王静,李健. 基于U-Net的高分辨率遥感图像土地利用信息提取[J]. 遥感技术与应用, 2021, 36(2): 285-292.
[4] 马彩虹,关琳琳,陈甫,王大成,刘建波. 基于内容的遥感图像变化信息检索概念模型设计[J]. 遥感技术与应用, 2020, 35(3): 685-693.
[5] 王宇,杨艺,王宝山,王田,卜旭辉,王传云. 深度残差神经网络高分辨率遥感图像建筑物分割[J]. 遥感技术与应用, 2019, 34(4): 736-747.
[6] 秦振涛,杨茹. 基于结构性字典学习的毛儿盖遥感图像去噪研究[J]. 遥感技术与应用, 2019, 34(4): 793-798.
[7] 秦振涛,杨茹,张靖,杨武年. 基于聚类结构自适应稀疏表示的高光谱遥感图像修复研究[J]. 遥感技术与应用, 2018, 33(2): 212-215.
[8] 王俊,秦其明,叶昕,王建华,秦雪彬,杨绣丞. 高分辨率光学遥感图像建筑物提取研究进展[J]. 遥感技术与应用, 2016, 31(4): 653-662.
[9] 张从梅,孙权森,王超,封磊,顾一禾. 基于非局部自相似性的遥感图像稀疏去噪算法[J]. 遥感技术与应用, 2016, 31(4): 739-747.
[10] 赵永光,李传荣,马灵玲,唐伶俐,王宁. 一种遥感图像太阳—观测几何归一化方法[J]. 遥感技术与应用, 2016, 31(2): 260-266.
[11] 肖新耀,许宁,尤红建. 一种基于à trous小波和联合稀疏表示的遥感图像融合方法[J]. 遥感技术与应用, 2015, 30(5): 1021-1026.
[12] 苏腾飞,李洪玉. 一种两阶段区域生长的遥感图像分割算法[J]. 遥感技术与应用, 2015, 30(3): 476-485.
[13] 侯鹏洋,季艳,高峰,胡蕾. 一种基于SIFT特征的快速逐层遥感图像配准方法[J]. 遥感技术与应用, 2014, 29(5): 873-877.
[14] 万智萍. 结合方向小波的多光谱与全色遥感图像融合算法[J]. 遥感技术与应用, 2014, 29(4): 660-668.
[15] 李青松,覃锡忠,贾振红,杨杰,胡英杰. 基于非下采样Shearlet和几何结构的遥感图像无监督变化检测[J]. 遥感技术与应用, 2014, 29(3): 482-488.