Please wait a minute...
img

官方微信

遥感技术与应用  2020, Vol. 35 Issue (3): 673-684    DOI: 10.11873/j.issn.1004-0323.2020.3.0673
数据与图像处理     
基于改进DenseNet网络的多源遥感影像露天开采区智能提取方法
张峰极1(),吴艳兰1,2(),姚雪东1,梁泽毓1
1.安徽大学资源与环境工程学院,安徽 合肥 230601
2.安徽省地理信息智能技术工程研究中心,安徽 合肥 230000
Opencast Mining Area Intelligent Extraction Method for Multi-source Remote Sensing Image based on Improved DenseNet
Fengji Zhang1(),Yanlan Wu1,2(),Xuedong Yao1,Zeyu Liang1
1.School of Resources and Environmental Engineering, Anhui University, Hefei 230601, China
2.Anhui Geographic Information Intelligent Technology Engineering Research Center, Hefei 230000, China
 全文: PDF(6688 KB)   HTML
摘要:

利用遥感技术对露天开采区进行信息提取和监测已成为解决矿山自然环境问题的重要手段。通过改进带密集连接的全卷积神经网络,构建露天开采区样本库,并训练了针对多源遥感数据的露天开采区提取模型,最终实现对铜陵地区露天开采区的全自动提取。与传统分类方法和深度学习方法相比,该方法在基于像元和基于对象的评价方面具有较好的精度,其中像元精度PA:0.977,交并比IoU:0.721,综合评价指标F1:0.838,Kappa系数:0.825,召回率:0.913,漏警率:0.087,虚警率:0.533。同时,该模型对于匀色较差的GoogleEarth影像也有较好的提取效果,表现出较强的泛化性和适用性,在多源遥感影像露天开采区提取方面具有较强的应用价值。

关键词: 深度学习全卷积神经网络DenseNet露天开采区提取全自动化    
Abstract:

The use of remote sensing technology for information extraction and monitoring of open-pit mining areas has become an important means to solve the natural environment problems of mines. Firstly, this paper improves the fully convolutional neural network with dense block. Then, the opencast mining area sample library is constructed, and the open-pit mining area extraction model for multi-source remote sensing data is trained. Finally, the automatic extraction of the opencast mining area is realized in Tongling. The results show that compared with traditional classification methods and deep learning methods, the proposed method has better accuracy in pixel-based and object-based evaluation. Specifically, the Pixel Accuracy (PA), Intersection over Union (IoU), F1, Kappa Coefficient, Recall, Missing Alarm and False Alarm is 0.977, 0.721, 0.838, 0.825, 0.913, 0.087 and 0.533, respectively. The model also has a great extraction effect for Google-Earth images with poor homogeneity, showing strong generalization and applicability. Therefore, the proposed model of this paper has wide application value in the extraction of opencast mining area by using multi-source remote sensing images.

Key words: Deep learning    Fully-Convolutional Neural Network    DenseNet    Opencast mining extraction    Fully automation
收稿日期: 2019-01-18 出版日期: 2020-07-10
ZTFLH:  TP79  
基金资助: 国家自然科学基金项目(41271445);安徽省自然科学基金项目(1308085MD52)
通讯作者: 吴艳兰     E-mail: 546598151@qq.com;wuyanlan@ahu.edu.cn
作者简介: 张峰极(1993-),男,安徽合肥人,硕士研究生,主要从事深度学习遥感影像信息提取方面的研究。E?mail:546598151@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  
张峰极
吴艳兰
姚雪东
梁泽毓

引用本文:

张峰极,吴艳兰,姚雪东,梁泽毓. 基于改进DenseNet网络的多源遥感影像露天开采区智能提取方法[J]. 遥感技术与应用, 2020, 35(3): 673-684.

Fengji Zhang,Yanlan Wu,Xuedong Yao,Zeyu Liang. Opencast Mining Area Intelligent Extraction Method for Multi-source Remote Sensing Image based on Improved DenseNet. Remote Sensing Technology and Application, 2020, 35(3): 673-684.

链接本文:

http://www.rsta.ac.cn/CN/10.11873/j.issn.1004-0323.2020.3.0673        http://www.rsta.ac.cn/CN/Y2020/V35/I3/673

图1  改进后的DenseNet网络
卫星遥感影像数据分辨率/m传感器时间/年数量/景用途
全色多光谱
高分一号28WFV/PMS2013~201412样本制作
高分二号14PMS20156样本制作
1模型测试
GoogleEarth影像/2CCD2014~201710样本制作
2015、20172模型测试
表1  不同传感器卫星遥感影像参数
图2  露天开采区样本库及栅格标签样本格式
图3  训练测试精度和损失变化曲线图
评价指标公式意义参数说明

①像元精度 (Pixel Accuracy)PA=i=0kpiji=0kj=0kpij值越大表示预测与真实值的像元匹配程度越高pii:正确提取的像元数量;pij和pji:错误提取的像元数量
②交并比(IoU)IoU=i=0kpijj=0kpij+j=0kpji-pii评估模型性能的标准指标
③综合评价指标(F1)

F1=2*precision*recallprecision+recall

precision=tptp+fp,recall=tptp+fn

衡量二分类模型精度的一种指标,它的最大值是1,最小值是0

tp:正确提取的像元个数;fn:漏提的像元个数

fp:错误提取的像元个数

④Kappa系数Kappa=p0-pe1-pe表示提取结果和真实值之间的吻合程度po和pc:分别代表每一类正确和错误提取的样本数量之和除以总样本数量
基于对象的评价方式⑤召回率(Recall)Recall=TPTP+FN正确提取个数(TP)与真实目标个数(TP+FN)的比值

TP:正确提取的个数;FN:漏提的个数;

FP:错误提取的个数

⑥漏警率 (MissingAlarm)MissingAlarm=FNTP+FN漏提出来的目标个数(TP)与真实目标个数(TP+FN)的比值
⑦虚警率 (FalseAlarm)FalseAlarm=FPTP+FP错误提取的个数(FP)与提取目标总个数(TP+FP)的比值
表2  本文采用的精度评价指标
图4  不同方法提取结果比较(铜陵地区露天开采区)
图5  不同深度学习模型提取结果比较(铜陵地区露天开采区)
像元精度(PA)交并比(IoU)综合评价指标(F1)Kappa系数
本文方法0.9770.7210.8380.825
最大似然法0.9450.5410.7020.673
支持向量机0.9620.5560.7150.695
决策树分类0.9540.3670.5370.518
DeepLab0.9670.5680.7250.708
U-Net0.9630.6150.7610.742
SegNet0.9790.7190.8370.826
表3  基于像元的精度评价

真实个数

(TP+FN)

提取个数

(TP+FP)

正确数(TP)漏警数(FN)虚警数(FP)

召回率

(Recall)

漏警率 (MissingAlarm)虚警率 (FalseAlarm)
本文方法2345212240.9130.0870.533
最大似然法2379203590.8700.1300.747
支持向量机2345176280.7390.2610.622
决策树分类2316617100.2610.7390.625
DeepLab2335176180.7390.2610.514
U-Net2364203440.8700.1300.687
SegNet232618580.7830.2170.308
表4  基于对象的精度评价
本文方法最大似然法支持向量机决策树分类
效率3 min10 min超过2 h超过2 h
自动化程度自动化程度高,不需要人工干预需要人工选取样本需要人工选取样本需要人工建树,确定分类规则
表5  效率和自动化对比
图6  GoogleEarth遥感影像露天开采区提取结果
图7  导致露天开采区提取产生误检的可能因素分析
1 Xue Weining. Monitoring and Application of Remote Sensing Technology in Mine Environmental Improvement and Restoration Project in Taishan Region[J]. Shandong Land and Resources, 2019,35(10):59-63.
1 薛卫宁. 遥感技术在泰山区域矿山环境治理修复工程的监测应用[J]. 山东国土资源, 2019,35(10):59-63.
2 Wang Jie, Yin Yaqiu, Yu Hang, et al. Remote Sensing Monitoring of Mine Geological Environment in Zhejiang Province based on RS and GIS[J]. Remote Sensing for Land & Resources, 2020,32(1):232-236.
2 汪洁, 殷亚秋, 于航,等. 基于RS和GIS的浙江省矿山地质环境遥感监测[J]. 国土资源遥感, 2020,32(1):232-236.
3 Xiang Jie, Chen JianPing, Li Shi, et al. Research on Application of Unmanned Aerial Vehicle Technology to Dynamic Monitoring of Reserves in the Shouyun Iron Mine, Beijing[J]. Remote Sensing for Land and Resources, 2018, 30(3) : 224-229.
3 向杰,陈建平,李诗,等.无人机遥感技术在北京首云铁矿储量动态监测中的应用[J].国土资源遥感,2018,30(3): 224-229.
4 Hu Yihong. Research of Remote Sensing Image in Rare Earth Mining by Satellite Remote Sensing Image[J]. Geology of Fujian, 2018, 37(1): 44-50.
4 胡一鸿.卫星遥感影像在稀土矿山动态监测研究[J].福建地质,2018,37(1):44-50.
5 Yu Hairuo, Yan Qin, Dong Chun, et al. Extraction of Features based on The Decision Tree Classification and Analysis of Pollution in Datun Count[J]. Geomatics and Spatial Information Technology, 2016, 39(4): 67-69, 72.
5 于海若,燕琴,董春,等.基于决策树分类的大屯矿区地物信息提取及矿区污染分析[J].测绘与空间地理信息,2016,39(4):67-69, 72.
6 Wang Lina, Chen Lu. Application Research of Support Vector Machine (SVM) Combined with Object-oriented Method in Information Extraction of Open-pit Mining Areas[C]∥ Academic Exchange Meeting of Surveying and Mapping Society of Six Provinces and One City in East China and Cross-Strait Surveying and Mapping Technology Exchange and Symposium, 2017:208-213.
6 王莉娜,程璐.支持向量机(SVM)结合面向对象法在露天矿区信息提取中的应用研究[C]//华东六省一市测绘学会学术交流会暨海峡两岸测绘技术交流与学术研讨会, 2017:208-213.
7 Li Shiping, Wu Wenbo. Artificial Neural Network and Its Application in Remote Sensing Image Processing[J]. Mine Surveying, 2007(3) : 28-30.
7 李世平,武文波.人工神经网络及其在遥感图像处理中的应用[J].矿山测量,2007(3):28-30.
8 Xie Jing, Wang Zongming, Mao Dehua, et al. Remote Sensing Classification of Wetlands Using Object-oriented Method and Multi-season HJ-1 Images: A Case Study in The Sanjiang Plain North of The Wandashan Mountain[J]. Wetland Science, 2012, 10(4) : 429-438.
8 谢静,王宗明,毛德华,等.基于面向对象方法和多时相HJ-1影像的湿地遥感分类——以完达山以北三江平原为例[J].湿地科学,2012,10(4):429-438.
9 Chen G, Hay G J, Carvalho L M T, et al. Object-based Change Detection[J]. International Journal of Remote Sensing, 2012, 33(14) : 4434-4457.
10 Desclée B, Bogaert P, Defourny P. Forest Change Detection by Statistical Object-based Method[J]. Remote Sensing of Environment,2006,102(1-2):1-11. doi:10.1016/j.rse. 2006. 01.013.
doi: 10.1016/j.rse. 2006. 01.013
11 Zu Qi, Yuan Xiping, Mo Yuanfu, et al. Surface Features' Information Extraction from SPOT Images with Object-oriented Classification Method[J]. Carsologica Sinica, 2011, 30(2) : 227-232.
11 祖琪,袁希平,莫源富,等.基于面向对象分类方法在SPOT影像中的地物信息提取[J].中国岩溶,2011,30(2):227-232.
12 Dai Jingjing, Wu Ya'nan, Wang Denghong, et al. Object-oriented Classification for The Extraction of Remote Sensing Information in Rare Earth Mining Areas[J]. Acta Geoscientica Sinica, 2018,39(1):111-118.
12 代晶晶, 吴亚楠, 王登红,等. 基于面向对象分类的稀土开采区遥感信息提取方法研究[J]. 地球学报, 2018,39(1):111-118.
13 Huang B, Zhao B, Song Y. Urban Land-Use Mapping Using a Deep Convolutional Neural Network with High Spatial Resolution Multispectral Remote Sensing Imagery[J]. Remote Sensing of Environment,2018,214:73-86.doi:https:
doi: https:
14 Zhang C, Sargent I, Pan X, et al. An Object-based Convolutional Neural Network (OCNN) for Urban Land Use Classification[J]. Remote Sensing of Environment, 2018,216: 57-70.doi:10.1016/j.rse.2018.06.034.
doi: 10.1016/j.rse.2018.06.034
15 Xu G, Zhu X,Tapper N. Using Convolutional Neural Networks Incorporating Hierarchical Active Learning for Target-searching in Large-scale Remote Sensing Images[J]. International Journal of Remote Sensing,2020, 41(11): 4057-4079. doi:10.1080/01431161.2020.1714774.
doi: 10.1080/01431161.2020.1714774
16 Mehra A, Jain N, Srivastava H S. A Novel Approach to Use Semantic Segmentation based Deep Learning Networks to Classify Multi-Temporal SAR Data[C]∥ Geocarto International, 2020:1-16. doi:10.1080/10106049.2019.1704072.
doi: 10.1080/10106049.2019.1704072
17 Wang Y, Li H, Jia P, et al. Multi-Scale DenseNets-based Aircraft Detection from Remote Sensing Images[J]. Sensors2019, 19, 5270. doi:10.3390/s19235270.
doi: 10.3390/s19235270
18 Zhang L, Zhang L, Du B. Deep Learning for Remote Sensing Data: A Technical Tutorial on the State of the Art[J]. IEEE Geoscience and Remote Sensing Magazine, 2016, 4: 22-40. doi:10.1109/MGRS.2016.2540798.
doi: 10.1109/MGRS.2016.2540798
19 Chen X, Xiang S, Liu C, et al. Vehicle Detection in Satellite Images by Hybrid Deep Convolutional Neural Networks[J]. IEEE Geoscience and Remote Sensing Letters2014, 11: 1797-1801. doi:10.1109/LGRS.2014.2309695.
doi: 10.1109/LGRS.2014.2309695
20 Wang Xue, Lichun Sui, Zhong Mianqing, et al. Fully Convolution Neural Networks for Water Extraction of Remote Sensing Images[J]. Bulletin of Surveying and Mapping, 2018(6): 41-45.王雪,隋立春,钟棉卿,等.全卷积神经网络用于遥感影像水体提取[J].测绘通报,2018(6):41-45.
21 Pei Liang, Liu Yang, Tan Hai, et al. Cloud Detection based on Improved Fully Convolutional Networks of ZY-3 Satellite Remote Sensing Images[J]. Laser and Optoelectronics Progress,2019,56(8):052801.doi:10.3788/LOP56.052801.
doi: 10.3788/LOP56.052801
21 [裴亮,刘阳,谭海,等.基于改进的全卷积神经网络的资源三号遥感影像云检测[J].激光与光电子学进展,2019,56(8):052801.doi:10.3788/LOP56.052801.
doi: 10.3788/LOP56.052801
22 Xu Y Y, Wu L, Xie Z, et al. Building Extraction in Very High Resolution Remote Sensing Imagery Using Deep Learning and Guided Filters[J]. Remote Sensing, 2018, 10, 144. doi:10.3390/rs10010144.
doi: 10.3390/rs10010144
23 Fu G, Liu C J, Zhou R, et al. Classification for High Resolution Remote Sensing Imagery Using a Fully Convolutional Network[J]. Remote Sensing, 2017, 9489. doi:10.3390/rs9050498.
doi: 10.3390/rs9050498
24 Fang Xu, Wang Guanghui, Yang Huachao, et al. High Resolution Remote Sensing Image Classification Combining with Mean-Shift Segmentation and Fully Convolution Neural Network[J]. Laser and Optoelectronics Progress, 2018,55(2):446-454.
24 方旭,王光辉,杨化超,等.结合均值漂移分割与全卷积神经网络的高分辨遥感影像分类[J].激光与光电子学进展, 2018,55(2):446-454.
25 Wang S, Yang H, Wu Q, et al. An Improved Method for Road Extraction from High-resolution Remote Sensing Images that Enhances Boundary Information[J]. Sensors, 2020, 20: 2064. doi:10.3390/s20072064.
doi: 10.3390/s20072064
26 Venugopal N. Automatic Semantic Segmentation with DeepLab Dilated Learning Network for Change Detection in Remote Sensing Images[J]. Neural Processing Letters, doi:10.1007/s11063-019-10174-x.
doi: 10.1007/s11063-019-10174-x
27 Xiong W, Xiong Z, Cui Y, et al. Deep Multi-feature Fusion Network for Remote Sensing Images[J]. Remote Sensing Letters, 2020,11(6):563-571. doi:10.1080/2150704X.2020. 1743376.
doi: 10.1080/2150704X.2020. 1743376
28 Marmanis D, Schindler K, Wegner J D, et al. Classification with an Edge: Improving Semantic with Boundary Detection[J]. ISPRS Journal of Photogrammetry and Remote Sensing2018, 135: 158-172. doi:10.1016/j.isprsjprs.2017.11.009.
doi: 10.1016/j.isprsjprs.2017.11.009
29 Audebert N, Le Saux B,Lefevre S, et al. Very High Resolution Urban Remote Sensing with Multimodal Deep Networks[J]. ISPRS Journal of Photogrammetry and Remote Sensing,2018, 140: 20-32. doi:10.1016/j.isprsjprs.2017.11.011.
doi: 10.1016/j.isprsjprs.2017.11.011
30 Li L, Liang J, Weng M, et al. A Multiple-feature Reuse Network to Extract Buildings from Remote Sensing Imagery[J]. Remote Sensing, 2018, 10: 1350. doi:10.3390/rs10091350.
doi: 10.3390/rs10091350
31 Huang G L, van der Maaten Z, Weinberger L, et al. Densely Connected Convolutional Network[R]. CoRR 2016, abs/1608.06993. doi:10.1109/CVPR.2017.243.
32 Simon Jégou M D, David V, Adriana R, et al. In The One Hundred Layers Tiramisu: Fully Convolutional Densenets for Semantic Segmentation[C]∥ IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, Hawaii, USA, 2017; 1175-1183. doi:10.1109/CVPRW. 2017.156.
doi: 10.1109/CVPRW. 2017.156
33 Xu Y Y, Xie Z, Feng Y X, et al. Road Extraction from High-Resolution Remote Sensing Imagery Using Deep Learning[J]. Remote Sensing, 2018, 10: 1461. doi:10.3390/rs10091461.
doi: 10.3390/rs10091461
34 Yang H, Wu P H, Yao X D, et al. Building Extraction in Very High Resolution Remote Sensing Imagery by Dense-Attention Networks[J]. Remote Sensing, 2018, 10(11):1768. doi:10.3390/rs10111768.
doi: 10.3390/rs10111768
35 The First National Geographic Survey of the National Geographic Survey of the State Council.The First National Geographic Survey of National Conditional and Surveys of Geographical Conditions and Indicatiors[S].Beijing:Surveying and Mapping Press,2013.
35 国务院第一次全国地理国情普查领导小组办公室.第一次全国地理国情普查地理国情普查内容与指标[S].北京:测绘出版社,2013.
36 Lee D H, Lee K M, Lee S U. Fusion of LiDAR and Imagery for Reliable Building Extraction[J]. Photogrammetric Engineering & Remote Sensing, 2008, 74(2):215-225. doi:10.14358/PERS.74.2.215.
doi: 10.14358/PERS.74.2.215
37 Gavankar N L, Ghosh S K. Automatic Building Footprint Extraction from High-resolution Satellite Image Using Mathematical Morphology[J]. European Journal of Remote Sensing, 2018, 51(1):182-193. doi:10.1080/22797254. 2017.1416676.
doi: 10.1080/22797254. 2017.1416676
38 Zhan Q M, Molenaar M, Tempfli K, et al. Quality Assessment for Geo-spatial Objects Derived from Remotely Sensed Data[J]. International Journal of Remote Sensing, 2005, 26(14):22. doi:10.1080/01431160500057764.
doi: 10.1080/01431160500057764
39 Rottensteiner F, Trinder J, Clode S, et al. Using the Dempster–Shafer Method for The Fusion of LiDAR Data and Multi-Spectral Images for Building Detection[J]. Information Fusion, 2005, 6(4):283-300.
40 Song W, Haithcoat T L. Development of Comprehensive Accuracy Assessment Indexes for Building Footprint Extraction[J]. IEEE Transactions on Geoscience and Remote Sensing, 2005, 43(2):402-404. doi:10.1109/tgrs.2004.838418.
doi: 10.1109/tgrs.2004.838418
41 Foody G M. Status of Land Cover Classification Accuracy Assessment[J]. Remote Sensing of Environment, 2002, 80(1):185-201. doi:10.1016/s0034-4257(01)00295-4.
doi: 10.1016/s0034-4257(01)00295-4
42 Rutzinger M, Rottensteiner F, Pfeifer N. A Comparison of Evaluation Techniques for Building Extraction from Airborne Laser Scanning[J]. IEEE Journal of Selected Topics in Applied Earth Observations & Remote Sensing, 2009, 2(1):11-20. doi:10.1109/jstars.2009.2012488.
doi: 10.1109/jstars.2009.2012488
43 Awrangjeb M, Fraser C S. An Automatic and Threshold-free Performance Evaluation System for Building Extraction Techniques from Airborne LiDAR Data[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2014, 7(10):4184-4198. doi:10.1109/JSTARS. 2014.2318694.
doi: 10.1109/JSTARS. 2014.2318694
[1] 李越帅,郑宏伟,罗格平,杨辽,王伟胜,桂东伟. 集成U-Net方法的无人机影像胡杨树冠提取和计数[J]. 遥感技术与应用, 2019, 34(5): 939-949.
[2] 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.
[3] 王颖洁,张荞,张艳梅,蒙印,郭文. 基于深度卷积神经网络的油罐目标检测研究[J]. 遥感技术与应用, 2019, 34(4): 727-735.
[4] 林志玮,丁启禄,黄嘉航,涂伟豪,胡典,刘金福. 基于DenseNet的无人机光学图像树种分类研究[J]. 遥感技术与应用, 2019, 34(4): 704-711.
[5] 刘天福,陈学泓,董琪,曹鑫,陈晋. 深度学习在GlobeLand30-2010产品分类精度优化中应用研究[J]. 遥感技术与应用, 2019, 34(4): 685-693.
[6] 徐梦竹, 徐佳, 邓鸿儒, 袁春琦. 基于全极化SAR影像的海岛地物分类[J]. 遥感技术与应用, 2019, 34(3): 647-654.
[7] 李淑敏, 冯权泷, 梁其椿, 张学庆. 基于深度学习的国产高分遥感影像飞机目标自动检测[J]. 遥感技术与应用, 2018, 33(6): 1095-1102.
[8] 何海清,庞燕,陈晓勇. 面向遥感影像场景的深度卷积神经网络递归识别模型[J]. 遥感技术与应用, 2017, 32(6): 1078-1082.