AED-Net：滑坡灾害遥感影像语义分割模型

doi:10.12082/dqxxkx.2023.230171

地球信息科学学报 ›› 2023, Vol. 25 ›› Issue (10): 2012-2025.doi: 10.12082/dqxxkx.2023.230171

• 地球信息科学理论与方法 • 上一篇下一篇

AED-Net：滑坡灾害遥感影像语义分割模型

蒋伟杰(), 张春菊^*(), 徐兵, 罗晨晨, 周晗, 周康

合肥工业大学土木与水利工程学院，合肥 230009

收稿日期:2023-04-03 修回日期:2023-06-01 出版日期:2023-10-25 发布日期:2023-09-22
通讯作者: * 张春菊（1984—），女，安徽宿州人，博士，副教授，主要从事地理信息智能处理与服务研究。 E-mail: zcjtwz@sina.com
作者简介:蒋伟杰（1998—），男，安徽阜阳人，硕士生，主要从事遥感影像智能处理与研究。E-mail: jwj_1219@163.com
基金资助:
国家自然科学基金项目(42171453);国家重点研发计划项目(2022YFB3904200)

AED-Net: Semantic Segmentation Model for Landslide Recognition from Remote Sensing Images

JIANG Weijie(), ZAHNG Chunju^*(), XU Bing, LUO Chenchen, ZHOU Han, ZHOU Kang

School of Civil Engineering, Hefei University of Technology, Hefei 230009, China

Received:2023-04-03 Revised:2023-06-01 Online:2023-10-25 Published:2023-09-22
Contact: * ZHANG Chunju, E-mail: zcjtwz@163.com
Supported by:
National Natural Science Foundation of China(42171453);National Key Research and Development Program of China(2022YFB3904200)

摘要/Abstract

摘要：

遥感影像蕴含丰富的语义信息，在滑坡灾害监测任务中发挥出了重要的作用。传统的滑坡识别主要通过遥感目视解译和人机交互识别，存在耗时费力、主观性强和提取精度低等问题。语义分割作为深度学习中的一项重要任务，因其端到端的像素级分类能力，已在遥感影像自动化识别任务中发挥出了重要作用。现有遥感影像滑坡灾害语义分割模型通常无法顾及多尺度地物特征，且随着网络深度增加会造成边界模糊等问题。本文提出了AED-Net（Attention combined with Encoder-Decoder Network），使用浅层特征提取网络缓解深度神经网络造成的边界模糊问题，利用空洞空间卷积池化金字塔结构的多尺度特征提取能力，结合编码器-解码器结构的特征还原能力还原边界信息，并使用通道注意力机制强化模型的关键特征学习能力。利用GID-5数据集针对模型中空洞卷积的膨胀率设置、通道注意力机制的选择进行对比试验以得到最优解，最终得到的模型在毕节市滑坡灾害数据集上获得了最优表现，像素准确度为95.58%，平均像素精度为89.24%，平均交互比为82.68%，相比PSP-Net、Attention U-Net、加入ECA注意力机制的DeeplabV3+、PA-Fov、LandsNet等语义分割模型，PA提升了0.73%~1.97%，MPA提升了1.0%~2.84%，MIoU提升了2.25%~5.11%，达到了最优分割效果。

关键词: 深度学习, 滑坡, 遥感影像, 编码器-解码器, 特征增强, 多尺度特征, 语义分割, 注意力机制

Abstract:

Remote sensing images contain rich semantic information and play an important role in landslide disaster monitoring. Traditional landslide recognition is mainly based on remote sensing visual interpretation and human-computer interaction recognition, which is time and labor consuming, with strong subjectivity and low extraction accuracy. Semantic segmentation, as an important task in deep learning, has played an important role in automatic recognition tasks using remote sensing images due to its end-to-end, pixel-level classification capability and has great potential in automatic recognition of landslides. The existing semantic segmentation models for landslides using remote sensing images usually lack the feature information of multi-scale ground objects, and the boundary will be blurred with the increase of network depth. In this paper, Attention combined with Encoder-Decoder Network (AED-Net) is proposed for landslide recognition. A shallow feature extraction network is used to alleviate the boundary ambiguity caused by deep neural network. Multi-scale feature extraction capability of convolution pool pyramid structure in void space is utilized. Combined with the feature restoring ability of the encoder-decoder structure, the boundary information is restored, and the channel attention mechanism is used to enhance the key feature learning ability of the model. The focal-loss function is used to alleviate the imbalance of positive and negative samples. In our study, firstly, the GID-5 data set is used to conduct comparative tests on the expansion rate setting of void convolutions and the selection of channel attention mechanism in the model to get the optimal solution. Then, the feature weight is transferred to the semantic segmentation task for landslide disaster by using transfer learning method, and the hyperparameter discussion and ablation experiment are carried out. The resulting model achieves the optimal segmentation performance on the landslide disaster data set of Bijie City, with a Pixel Accuracy (PA) of 95.58%, the Mean Pixel Accuracy (MPA) of 89.24%, and the Mean Intersection over Union (MIoU) of 82.68%. Compared with classical semantic segmentation networks such as PSP-Net, Attention U-Net, DeeplabV3+ with ECA attention mechanism, and semantic segmentation models such as PA-Fov and LandsNet for classfifying landslide disasters, the pixel accuracy of our model increases by 0.73%~1.97%. The average pixel accuracy of all categories increases by 1.0%~2.84%, and the average interaction ratio increases by 2.25%~5.11%. Moreover, the edge information of landslide image is smoother and the multi-scale landslide segmentation accuracy is better than other deep learning models, which demonstrates the effectiveness of the proposed model in semantic segmentation of landslides from remote sensing images.

Key words: deep learning, landslide, remote sensing image, encoder-decoder, feature enhancement, multi-scale characteristics, semantic segmentation, attention mechanism

蒋伟杰, 张春菊, 徐兵, 罗晨晨, 周晗, 周康. AED-Net：滑坡灾害遥感影像语义分割模型[J]. 地球信息科学学报, 2023, 25(10): 2012-2025.DOI:10.12082/dqxxkx.2023.230171

JIANG Weijie, ZAHNG Chunju, XU Bing, LUO Chenchen, ZHOU Han, ZHOU Kang. AED-Net: Semantic Segmentation Model for Landslide Recognition from Remote Sensing Images[J]. Journal of Geo-information Science, 2023, 25(10): 2012-2025.DOI:10.12082/dqxxkx.2023.230171

图/表 17

图1

图2

图3

图4

图5

图6

图7

表1

表2

表3

表4

表5

表6

表7

图8

图9

图10

参考文献 29

[1]	许嘉慧, 孙德亮, 张虹. 多尺度滑坡灾害生态风险评价与风险管理研究──以三峡库区万州-巫山段为例[J/OL]. 生态学报, 2023(11):1-10.
	[Xu J H, Sun D L, Zhang H, et al. Multi-scale ecological risk assessment and risk management of landslide disasters: a case study of Wanzhou-Wushan section in the Three Gorges Reservoir area. Acta Ecologica Sinica, 2023, 43(11).] http://kns.cnki.net/kcms/detail/11.2031.q.20230207.1043.023.html.
[2]	许强, 董秀军, 李为乐. 基于天-空-地一体化的重大地质灾害隐患早期识别与监测预警[J]. 武汉大学学报·信息科学版, 2019, 44(7):957-966.
	[Xu Q, Dong X J, Li W L. Integrated space-air-ground early detection, monitoring and warning system for potential catastrophic geohazards[J]. Geomatics and Information Science of Wuhan University, 2019, 44(7):957-966.] DOI:10.13203/j.whugis20190088
[3]	Guzzetti F, Mondini A C, Cardinali M, et al. Landslide inventory maps: New tools for an old problem[J]. Earth-Science Reviews, 2012, 112(1/2):42-66. DOI:10.1016/j.earscirev.2012.02.001
[4]	Chen H X, Zhang S, Peng M, et al. A physically-based multi-hazard risk assessment platform for regional rainfall-induced slope failures and debris flows[J]. Engineering Geology, 2016, 203:15-29. DOI:10.1016/j.enggeo.2015.12.009
[5]	梅伟, 顾世祥, 刘鑫, 等. 基于滑坡大变形过程的滑坡定量风险评估方法[J]. 武汉大学学报·工学版, 2022, 55(5):443-453.
	[Mei W, Gu S X, Liu X, et al. Quantitative risk assessment of landslides based on large deformation process of landslide[J]. Engineering Journal of Wuhan University, 2022, 55(5):443-453.] DOI:10.14188/j.1671-8844.2022-05-003
[6]	Haneberg W C, Cole W F, Kasali G. High-resolution lidar-based landslide hazard mapping and modeling, UCSF Parnassus Campus, San Francisco, USA[J]. Bulletin of Engineering Geology and the Environment, 2009, 68(2):263-276. DOI:10.1007/s10064-009-0204-3
[7]	Xu C. Preparation of earthquake-triggered landslide inventory maps using remote sensing and GIS technologies: Principles and case studies[J]. Geoscience Frontiers, 2015, 6(6):825-836. DOI:10.1016/j.gsf.2014.03.004
[8]	胡芳驰, 樊雅婧. 遥感影像滑坡灾害识别应用研究[J]. 农业灾害研究, 2021, 11(4):98-99.
	[Hu F C, Fan Y J. Application of remote sensing image for landslide disaster identification[J]. Journal of Agricultural Catastrophology, 2021, 11(4):98-99.] DOI:10.3969/j.issn.2095-3305.2021.04.040
[9]	Shelhamer E, Long J, Darrell T. Fully convolutional networks for semantic segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(4):640-651. DOI:10.1109/TPAMI.2016.2572683 pmid: 27244717
[10]	Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for biomedical image segmentation[M]//Lecture Notes in Computer Science. Cham: Springer International Publishing, 2015:234-241. DOI:10.1007/978-3-319-24574-4_28
[11]	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. pmid: 28060704
[12]	Zhao H S, Shi J P, Qi X J, et al. Pyramid scene parsing network[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2017:6230-6239. DOI:10.1109/CVPR.2017.660
[13]	Chen L C, Zhu Y K, Papandreou G, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation[C]//European Conference on Computer Vision. Cham: Springer, 2018:833-851.10.1007/978-3-030-01234-2_49
[14]	Xie E Z, Wang W H, Yu Z D, et al. SegFormer: Simple and efficient design for semantic segmentation with transformers[EB/OL]. 2021: arXiv:2105.15203. https://arxiv.org/abs/2105.15203.
[15]	Cheng L B, Li J, Duan P, et al. A small attentional YOLO model for landslide detection from satellite remote sensing images[J]. Landslides, 2021, 18(8):2751-2765. DOI: 10.1007/s10346-021-01694-6
[16]	Ullo S L, Mohan A, Sebastianelli A, et al. A new mask R-CNN-based method for improved landslide detection[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2021, 14:3799-3810. DOI: 10.1109/JSTARS.2021.3064981
[17]	Bragagnolo L, Rezende L R, da Silva R V, et al. Convolutional neural networks applied to semanticsegmentation of landslide scars[J]. CATENA, 2021, 201:105189. DOI: 10.1016/j.catena.2021.105189
[18]	Mu F, Li J N, Shen N, et al. Pixel-adaptive field-of-view for remote sensing image segmentation[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19:1-5. DOI:10. 1109/LGRS.2022.3187049
[19]	Yi Y N, Zhang W C. A new deep-learning-based approach for earthquake-triggered landslide detection from single-temporal RapidEye satellite imagery[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2020, 13:6166-6176. DOI:10.1109/JSTARS.2020.3028855
[20]	Dosovitskiy A, Beyer L, Kolesnikov A, et al. An image is worth 16x16 words: Transformers for image recognition at scale[EB/OL]. 2020: arXiv: 2010.11929. https://arxiv.org/abs/2010.11929
[21]	Sandler M, Howard A, Zhu M L, et al. MobileNetV2: inverted residuals and linear bottlenecks[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. IEEE, 2018:4510-4520. DOI:10.1109/CVPR.2018.00474.
[22]	He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2016:770-778. DOI:10.1109/CVPR.2016.90
[23]	Hassanin M, Anwar S, Radwan I, et al. Visual attention methods in deep learning: An In-depth survey[EB/OL]. 2022:arXiv:2204.07756. https://arxiv.org/abs/2204.07756
[24]	Hu J, Shen L, Sun G. Squeeze-and-excitation networks[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. IEEE, 2018:7132-7141. DOI: 10.1109/CVPR.2018.00745
[25]	Woo S, Park J, Lee J Y, et al. CBAM: convolutional block attention module[C]//European Conference on Computer Vision. Cham: Springer, 2018:3-19.10.1007/978-3-030-01234-2_1
[26]	Wang Q L, Wu B G, Zhu P F, et al. ECA-net: Efficient channel attention for deep convolutional neural networks[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, 2020:11531-11539. DOI:10.1109/CVPR42600.2020.01155
[27]	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. DOI: 10.1109/TPAMI.2018.2858826
[28]	Ji S P, Yu D W, Shen C Y, et al. Landslide detection from an open satellite imagery and digital elevation model dataset using attention boosted convolutional neural networks[J]. Landslides, 2020, 17(6):1337-1352. DOI:10.1007/s10346-020-01353-2
[29]	Oktay O, Schlemper J, Le Folgoc L, et al. Attention U-net: Learning where to look for the pancreas[EB/OL]. 2018:arXiv:1804.03999. https://arxiv.org/abs/1804.03999

膨胀率	PA	MPA	MIOU
2	85.67	82.24	70.91
4	83.16	78.23	66.56
6	84.09	80.13	70.19
8	84.81	81.37	69.06

模型	PA	MPA	IoU
模型	PA	MPA	建筑用地	农业用地	森林	湿地	水体
Without-Attention	85.64	79.44	77.16	72.35	59.73	57.20	84.53
With-CBAM	85.06	82.21	75.55	71.0	56.61	63.79	84.37
With-SE	85.78	81.82	77.29	71.95	58.86	63.25	83.21
With-ECA	87.41	85.24	77.53	72.88	60.25	64.11	86.86

模型	F-Score
不使用迁移学习	0.842
迁移学习	0.886

Batch Size	PA	MPA	MIoU
2	94.44	81.43	79.84
4	94.45	84.4	80.75
8	95.47	88.57	81.56
16	95.58	89.24	82.68

学习率	PA	MPA	MIoU
1e-2	91.72	78.59	69.90
1e-3	95.11	89.68	81.55
1e-4	95.48	89.04	82.34
5e-5	95.17	89.42	81.63
1e-5	95.48	89.8	82.68

AED-Net：滑坡灾害遥感影像语义分割模型

AED-Net: Semantic Segmentation Model for Landslide Recognition from Remote Sensing Images

RichHTML

PDF (PC)

赞

可视化

摘要/Abstract

引用本文

使用本文

图/表 17

参考文献 29

相关文章 15

Metrics

本文评价

推荐阅读 0

	Focal-loss	Encoder-ECA	Decoder-ECA	MIoU/%
1	×	×	×	80.56
2	√	×	×	81.28
3	√	×	√	81.88
4	√	√	×	82.04
5	√	√	√	82.68

模型	PA	MPA	MIoU
PSP-Net^[12]	93.61	88.45	77.57
PA-FoV^[18]	94.26	86.4	78.37
ECA^[22]&DeeplabV3+^[13]	94.23	88.54	78.99
Attention U-Net^[29]	94.52	86.83	79.30
LandsNet^[19]	94.85	88.24	80.43
本文模型	95.58	89.24	82.68

[1]	周超, 甘露露, 王悦, 吴宏阳, 喻进, 曹颖, 殷坤龙. 综合非滑坡样本选取指数与异质集成机器学习的区域滑坡易发性建模[J]. 地球信息科学学报, 2023, 25(8): 1570-1585.
[2]	张俊瑶, 杨晓梅, 王志华, 杨海坤, 张博淳, 万庆, 雷梅. 绿色发展理念视角下内蒙古煤矿区格局演变分析[J]. 地球信息科学学报, 2023, 25(8): 1655-1668.
[3]	左溪冰, 刘智, 金飞, 林雨准, 王淑香, 刘潇, 李美霖. 面向高光谱影像小样本分类的全局-局部特征自适应融合方法[J]. 地球信息科学学报, 2023, 25(8): 1699-1716.
[4]	林娜, 何静, 王斌, 唐菲菲, 周俊宇, 郭江. 结合植被光谱特征与Sep-UNet的城市植被信息智能提取方法[J]. 地球信息科学学报, 2023, 25(8): 1717-1729.
[5]	张春菊, 刘文聪, 张雪英, 叶鹏, 汪陈, 朱少楠, 张达玉. 基于本体的金矿知识图谱构建方法[J]. 地球信息科学学报, 2023, 25(7): 1269-1281.
[6]	张彤, 刘仁宇, 王培晓, 高楚林, 刘杰, 王望舒. 感知物理先验的机器学习及其在地理空间智能中的研究前景[J]. 地球信息科学学报, 2023, 25(7): 1297-1311.
[7]	吴敏, 张明达, 李盼盼, 张勇健. 面向多源遥感影像数据的溯源模型研究[J]. 地球信息科学学报, 2023, 25(7): 1325-1335.
[8]	曹兴文, 吴孟泉, 郑雪婷, 郑宏伟, 李映祥, 张安安. 室内空间布局约束下的在线跟踪注册学习方法[J]. 地球信息科学学报, 2023, 25(7): 1418-1431.
[9]	柳林, 谢华芳, 岳瀚. 不同街景影像微环境提取方法对街面财产犯罪解释的差异分析[J]. 地球信息科学学报, 2023, 25(7): 1432-1447.
[10]	赵丹, 杜萍, 刘涛, 令振飞. 融合图自编码器与GRU的城市盗窃犯罪时空分布预测模型[J]. 地球信息科学学报, 2023, 25(7): 1448-1463.
[11]	金广垠, 沙恒宇, 张金雷, 黄金才. 基于路段-路口联合建模的对偶图卷积网络的行程时间估计方法[J]. 地球信息科学学报, 2023, 25(7): 1500-1513.
[12]	林炫歆, 肖桂荣, 周侯伯. 顾及土地利用动态变化的滑坡易发性评估方法[J]. 地球信息科学学报, 2023, 25(5): 953-966.
[13]	刘潇, 刘智, 林雨准, 王淑香, 左溪冰. 面向遥感影像场景分类的类中心知识蒸馏方法[J]. 地球信息科学学报, 2023, 25(5): 1050-1063.
[14]	侯慧太, 蓝朝桢, 徐青. 基于卫星影像全局和局部深度学习特征检索的无人机绝对定位方法[J]. 地球信息科学学报, 2023, 25(5): 1064-1074.
[15]	陈科, 管海燕, 雷相达, 曹爽. 基于特征增强核点卷积网络的多光谱LiDAR点云分类方法[J]. 地球信息科学学报, 2023, 25(5): 1075-1087.