Please wait a minute...
img

官方微信

遥感技术与应用  2023, Vol. 38 Issue (4): 990-1002    DOI: 10.11873/j.issn.1004-0323.2023.4.0990
遥感应用     
基于机器学习的棚户区识别应用——以上海棚户区为例
徐丹1(),林文鹏1,2(),马帅1
1.上海师范大学 环境与地理科学学院,上海 200234
2.上海长三角城市湿地生态系统国家野外科学观测研究站,上海 200234
Application on Slum Identification Using Machine Learning Methods: A Case Study of Shanghai Slums
Dan XU1(),Wenpeng LIN1,2(),Shuai MA1
1.School of Environmental and Geographical Sciences,Shanghai Normal University,Shanghai 200234,China
2.Yangtze River Delta Urban Wetland Ecosystem National Field Observation and Research Station,Shanghai 200234,China
 全文: PDF(8914 KB)   HTML
摘要:

精准获取并识别棚户区的空间分布及形态,对改善人居环境、优化城市空间结构具有重要意义。传统的实地调查方法耗时费力,以上海杨浦区南部的棚户区为例,从高空间分辨率影像中提取光谱、纹理和结构特征作为输入数据,提出了基于机器学习算法的高分遥感影像的棚户区提取方法。首先,综合比较最近邻(KNN)、逻辑回归(LR)、支持向量机(SVM)、随机森林(RF)、集成学习(EL)5种机器学习识别方法的适用性,确定最优分类器。其次,基于网格(50m×50m)对高分影像进行特征提取,并对特征网格分类、构建分类图像特征数据集。最后,通过图像特征进行棚户区识别,并评估5种机器学习算法在城市地区识别棚户区的能力及应用。结果表明:监督机器学习方法在提取棚户区的精度方面基本能够满足棚户区的研究及实际应用,但综合考量分类结果、分类精度和运行效率,EL算法Kappa系数为73.0%、总体精度为97.27%、召回率为79.02%,均高于其它算法,且漏分错误最少,能够更加完整、准确地完成棚户区信息提取。当考虑运行效率时,LR算法识别速度明显高于其它算法,更适用于大范围棚户区的使用需求。监督学习识别方法不仅可应用于高分影像特征识别,在遥感监测、城市规划和测绘等方面也具有较大的应用潜力。

关键词: 高分遥感影像棚户区机器学习支持向量机集成学习    
Abstract:

Accurately extracting and identifying the spatial distribution and form of slum is of great significance to improving the living environment and optimizing urban spatial structure. Traditional field investigation methods are time-consuming and laborious. As slums in the southern Yangpu District of Shanghai as the study area, spectral, textural and structural features from high-resolution remote sensing images as input data, this paper proposed an identification method for the slum using Machine Learning (ML) algorithms. Firstly, K-Nearest Neighbor (KNN), Logistic Regression(LR), Support Vector Machine(SVM), Random Forest (RF) and Ensemble Learning(EL) algorithms were compared comprehensively to determine the optimal classifier. Secondly, features were extracted from high-resolution images based on the grid of 50 m×50 m. Then the feature grid was classified and the feature dataset was constructed. Finally, the slum area was identified by image features, and the ability and application of five ML methods in urban area are evaluated. Results showed that supervised machine learning methods could basically meet the research and practical application of slums identification. In terms of the classification results, classification accuracy and operation efficiency, the Kappa coefficient of EL algorithm was 73.0%, the overall accuracy was 97.27%, and the recall rate was 79.02%, which were all higher than other algorithms, and the omission errors were the least. Therefore, the EL algorithm could complete the information extraction of slums more completely and accurately. When considering the operating efficiency, the LR algorithm had a higher identification speed than other algorithms and was more suitable for the use of a large range of slums. Moreover, ML methods could not only be used for features extracting in high-resolution images, but also had great application potential in remote sensing monitoring, urban planning and mapping.

Key words: High-resolution remote sensing image    Slum    Machine learning    Support vector machine    Ensemble learning
收稿日期: 2022-04-18 出版日期: 2023-09-11
ZTFLH:  TP75  
基金资助: 上海市自然科学基金项目(23ZR1446700);国家自然科学基金项目(41730642)
通讯作者: 林文鹏     E-mail: 1000497916@smail.shnu.edu.cn;linwenpeng@shnu.edu.cn
作者简介: 徐 丹(1993-),女,福建建阳人,博士研究生,主要从事环境演变与风险管理研究。E?mail: 1000497916@smail.shnu.edu.cn
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  
徐丹
林文鹏
马帅

引用本文:

徐丹,林文鹏,马帅. 基于机器学习的棚户区识别应用——以上海棚户区为例[J]. 遥感技术与应用, 2023, 38(4): 990-1002.

Dan XU,Wenpeng LIN,Shuai MA. Application on Slum Identification Using Machine Learning Methods: A Case Study of Shanghai Slums. Remote Sensing Technology and Application, 2023, 38(4): 990-1002.

链接本文:

http://www.rsta.ac.cn/CN/10.11873/j.issn.1004-0323.2023.4.0990        http://www.rsta.ac.cn/CN/Y2023/V38/I4/990

图1  研究区位置概况
图2  棚户区提取流程图
类型变量指标描述
光谱特征变量MEAN 1波段1的像素平均值
DEVST 1波段1的像素标准差
MAJORITY 1波段1的像素多数值
MEAN 2波段2的像素平均值
DEVST 2波段2的像素标准差
MAJORITY 2波段2的像素多数值
MEAN 3波段3的像素平均值
DEVST 3波段3的像素标准差
MAJORITY 3波段3的像素多数值
纹理特征变量MEAN_EDG边缘因子均值
DEVST_EDG边缘因子标准差
UNIFORGLCM的均匀性
ENTROPGLCM的商
CONTRASGLCM的对比
IDMGLCM的逆差距
COVARGLCM的协方差
VARIANGLCM的方差
CORRELACGLCM的相关性
SKEWNESS直方图的偏度
KURTOSIS直方图的峰度
结构特征变量RVF滞后一期比方差
RSF滞后一期比方差与滞后二期比方差之比
FDO原点附近的一阶导数
SDT滞后三期的二阶导数
MFM到第一个最大值的半变异函数平均值
VFM半方差函数值到第一个最大值的方差
DMF到第一个最大值的半变异函数平均值与滞后一期半方差的差值
RMM第一局部最大值的半方差与到这个最大值的半变异函数平均值的比率
SDF滞后一期与第一最大值之间的二阶分差
AFM滞后一期半变异函数值与到第一个最大值的半变异函数之间的面积
表1  特征指标
图3  GE图像规则网格中的城市街道和抽样区域
总计棚户区非棚户区训练集测试集
数据集8 3724647 9084 0003 772
表2  数据集的组成(网格数量)
True Positives (TP)False Positives (FP)
False Negatives (FN)True Negatives (TN)
表3  混淆矩阵
方法超参数
KNNn_neighbors: 8、weight: uniform
LRpenalty: L2、C、degree
SVMd: 3、gamma: 0.5
RFn_estimators: 500、max_features: 16、min_sample_leaf: 5
ELvoting: soft
表4  算法超参数设置
图4  各算法测试结果比较
图5  各算法提取棚户区局部效果对比图(R1-R4为图4标记区域)
方法OA/%

Precision

/%

Recall

/%

Kappa

/%

运行时间

/s

KNN96.6764.1976.2467.90.063 4
LR97.0367.4579.0171.20.072 3
SVMk96.5668.8757.4660.90.071 0
SVMrbk97.1479.3365.7570.60.092 8
RF96.8972.2665.0566.70.159 8
EL97.2770.4479.0273.03.608 4
表5  各算法的精度评价结果比较
1 LI Naisheng. Thoughts on prevention and control of urban shanty towns[J]. Urban Development Studies, 2000, 7(1):32-34.
1 李乃胜. 城市棚户区防治的思考[J]. 城市发展研究, 2000, 7(1):32-34.
2 WANG Mingfeng, CHENG Hong, NING Yuemin. Social integration of migrants in Shanghai's urban villages[J]. Acta Geographica Sinica, 2015, 70(8):1243-1254.
2 汪明峰, 程红, 宁越敏. 上海城中村外来人口的社会融合及其影响因素[J]. 地理学报, 2015, 70(8): 1243-1254.
3 Nations United. International development strategy for the third United Nations development decade[M]. New York: United Nations, 1980.
4 Nations United. Transforming our world: the 2030 Agenda for Sustainable Development[M]. New York: United Nations, 2015.
5 GAO Xiuxiu, ZHANG Xiaotong, HE Zheng. Emergence of the urban goal and its practice: Background and progress of the SDG11[J]. China Population,Resources and Environment, 2021, 31(11):144-154.
5 高秀秀, 张晓彤, 何正. 城镇和人类住区议题的演进与实践: SDG11的形成背景及执行进展[J], 中国人口·资源与环境, 2021, 31(11):144-154.
6 HUANG X, LIU H, ZHANG L. Spatiotemporal detection and analysis of urban villages in mega city regions of China using high-resolution remotely sensed imagery[J]. IEEE Transactions on Geoscience and Remote Sensing, 2015, 53(7):3639-3657. DOI: 10.1109/TGRS.2014.2380779
doi: 10.1109/TGRS.2014.2380779
7 KUFFER M, PFEFFER K, SLIUZAS R. Slums from space—15 years of slum mapping using remote sensing[J]. Remote Sensing, 2016, 8(6):455. DOI: 10.3390/rs8060455
doi: 10.3390/rs8060455
8 HOFMAN P, STROBL J, BLASCHLE T, et al. Detecting informal settlements from QuickBird data in Rio de Janeiro using an object based approach[J]. Springer Berlin Heidelberg, 2008.DOI: 10.1007/978-3-540-77058-9_29
doi: 10.1007/978-3-540-77058-9_29
9 SEBASTIAN D O, BODO C, BIRGIT K. An object-based classification approach for mapping migrant housing in the mega-urban area of the Pearl River Delta (China)[J].Remote Sensing, 2011, 3(8):1710-1723. DOI: 10.3390/rs3081710
doi: 10.3390/rs3081710
10 JIANG Dong, CHEN Shuai, DING Fangyu, et al. Classification of remote sensing image based on object-oriented method: A case study of Baixiang County[J]. Remote Sensing Technology and Application, 2018, 33(1):143-150.
10 江东, 陈帅, 丁方宇, 等. 基于面向对象的遥感影像分类研究——以河北省柏乡县为例[J]. 遥感技术与应用, 2018, 33(1):143-150.
11 ZHANG Liangpei, SHEN Huanfeng. Progress and future of remote sensing data fusion[J]. Journal of Remote Sensing,2016,20(5):1050-1061.
11 张良培, 沈焕锋. 遥感数据融合的进展与前瞻[J]. 遥感学报, 2016, 20(5):1050-1061.
12 ZHAO Yunhan, CHEN Gangqiang, CHEN Guangliang, et al. Integrating multi-source big data to extract buildings of urban villages: A case study of Tianhe district, Guangzhou[J]. Geography and Geo-Information Science,2018,34(5):7-13.
12 赵云涵,陈刚强, 陈广亮,等.耦合多源大数据提取城中村建筑物——以广州市天河区为例[J]. 地理与地理信息科学,2018,34(5):7-13.
13 CUI Cheng, ZHAO Lu, REN Hongyan,et al. Integrating high-resolution remote sensing image and street view image to identify urban village: A case study in Yuexiu district, Guangzhou city[J].Journal of Remote Sensing,2020,26(9):1802-1813.
13 崔成,赵璐,任红艳,等.耦合高分遥感影像与街景影像的广州市越秀区城中村识别[J]. 遥感学报,2020,26(9):1802-1813.
14 FENG Quanlong, CHEN Boan, NIU Bowen, et al. Identification of urban villages from remote sensing image based on multi-scale dilated convolutional neural network[J]. Transactions of the Chinese Society of Agricultural Machinery, 2021, 52(11):181-218.
14 冯权泷, 陈泊安, 牛博文, 等. 基于多尺度扩张卷积神经网络的城中村遥感识别[J]. 农业机械学报, 2021, 52(11):181-218.
15 WURM M, WEIGAND M, SCHMITT A, et al. Exploitation of textural and morphological image features in Sentinel-2A data for slum mapping[C]∥2017 Joint Urban Remote Sensing Event(JURSE),2017.Dubai,UAE:IEEE,2017:1-4.
16 LEONITA G, KUFFER M, SLIUZAS R, et al. Machine learning-based slum mapping in support of slum upgrading programs: The case of Bandung City,Indonesia[J].Remote sensing,2018,10(10):1522.DOI:10.3390/rs10101522
doi: 10.3390/rs10101522
17 DUQUE J C, PATINO J E, BETANCOURT A. Exploring the potential of machine learning for automatic slum identification from VHR imagery[J]. Remote Sensing,2017,9(9):895. DOI: 10.3390/rs9090895
doi: 10.3390/rs9090895
18 WEAVER J, MOORE B, REITH A, et al. A comparison of machine learning techniques to extract human settlements from high resolution imagery[C]//IGARSS 2018-2018 IEEE International Geoscience and Remote Sensing Symposium, 2018. Valencia, Spain: IEEE, 2018: 6412-6415.
19 LIAN Xihong, QI Yuan, WANG Hongwei, et al. Automatic extraction of residential information based on object-oriented in the Areas around the Qinghai Lake[J]. Remote Sensing Technology and Application, 2020, 35(4):775-785.
19 连喜红, 祁元, 王宏伟, 等. 基于面向对象的青海湖环湖区居民地信息自动化提取[J]. 遥感技术与应用, 2020, 35(4):775-785.
20 MENG Meijun. Research on Shanghai shanty area space change (1927-present)[D]. Shanghai: East China Normal University, China, 2006.孟眉军. 上海市棚户区空间变迁研究(1927年—至今)[D]. 上海:华东师范大学, 2006.
21 TAUBENBÖCK H, KRAFF N J. The physical face of slums: A structural comparison of slums in Mumbai, India, based on remotely sensed data[J]. Journal of Housing and the Built Environment, 2013, 29(1):15-38.
22 KIT O, LÜDEKE M. Automated detection of slum area change in Hyderabad, India using multitemporal satellite imagery[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2013, 83:130-137.
23 DUQUE J C, PATINO J E, RUIZ L A, et al. Measuring intra-urban poverty using land cover and texture metrics derived from remote sensing data[J].Landscape and Urban Planning, 2015, 135:11-21. DOI: 10.1016/j.landurbplan.2014.11.009
doi: 10.1016/j.landurbplan.2014.11.009
24 RUIZ L A, RECIO J A, FERNÁNDEZ-SARRÍA A, et al. A feature extraction software tool for agricultural object-based image analysis[J]. Computers and Electronics in Agriculture, 2011, 76(2):284-296. DOI:10.1016/j.compag.2011.02.007
doi: 10.1016/j.compag.2011.02.007
25 RHINANE H, HILALI A, BERRADA A, et al. Detecting slums from SPOT data in Casablanca Morocco using an object based approach[J]. Journal of Geographic Information System, 2011, 3(3):217-224. DOI: 10.4236/jgis.2011.33018
doi: 10.4236/jgis.2011.33018
26 YANG Lihua, DAI Qi, GUO Yanjun. Study on KNN text categorization algorithm[J]. Control and Automation, 2006, 22(21):269-270, 185.
26 杨丽华, 戴齐, 郭艳军. KNN文本分类算法研究[J]. 微计算机信息, 2006, 22(21):269-270, 185.
27 DEVIJVER P A, KITTLER J. Pattern recognition: A statistical approach[J]. Prentice-Hall International, 1982.
28 WANG Zhenfei, LIU Kaili, ZHENG Zhiyun,et al. Prediction retweeting of microblog based on logistic regression model[J]. Journal of Chinese Computer Systems,2016,37(8):1651-1655.
28 王振飞, 刘凯莉, 郑志蕴, 等. 基于逻辑回归模型的微博转发预测[J]. 小型微型计算机系统,2016,37(8):1651-1655.
29 LI Hang. Statistical learning methods[M]. Beijing: Tsinghua University Press, 2012.
29 李航. 统计学习方法[M]. 北京: 清华大学出版社, 2012.
30 ZHOU Z H. Machine learning[M]. Beijing: Tsinghua University Press, 2016.
30 周志华. 机器学习[M]. 北京: 清华大学出版社, 2016.
[1] 齐文栋,郑学昌,何黎明,卢珍,顾晓鹤,周艳兵. 基于多时相Sentinel-2影像的棉花雹灾时序变化遥感监测[J]. 遥感技术与应用, 2023, 38(3): 566-577.
[2] 范沛薇,郝蒙蒙,江东,丁方宇. 基于高分遥感影像的露天矿区土地利用动态监测——以平朔矿区为例[J]. 遥感技术与应用, 2023, 38(2): 274-284.
[3] 邴芳飞,金永涛,张文豪,徐娜,余涛,张丽丽,裴莹莹. 基于机器学习的遥感影像云检测研究进展[J]. 遥感技术与应用, 2023, 38(1): 129-142.
[4] 张妮娜,张珂,李运平,李曦,刘涛. 中国南方典型湿润山区植被类型的无人机多光谱遥感机器学习分类研究[J]. 遥感技术与应用, 2023, 38(1): 163-172.
[5] 刘志君,崔丽娟,李伟,窦志国,左雪燕,雷茵茹,潘旭,李晶,赵欣胜,翟夏杰. 基于高光谱的辽河口盐地碱蓬生态化学计量特征反演研究[J]. 遥感技术与应用, 2023, 38(1): 239-250.
[6] 高帅,侯学会,汪云,王倩,陈悦,邢瑞,王晶. 基于机器学习和大数据平台的陆地生态系统碳收支遥感监测[J]. 遥感技术与应用, 2022, 37(5): 1190-1197.
[7] 石育青,梁继,李云星,孟赛颖,石倩. 林地山区滑坡遥感的最优融合方法及应用[J]. 遥感技术与应用, 2022, 37(5): 1248-1258.
[8] 关士英,谢传节,袁占良,刘高焕. 基于SVM的高分时序冬小麦种植区提取研究[J]. 遥感技术与应用, 2022, 37(3): 629-637.
[9] 方攀飞,王雷光,徐伟恒,欧光龙,代沁伶,李若楠. 云南香格里拉地区森林优势树种决策融合分类[J]. 遥感技术与应用, 2022, 37(3): 638-650.
[10] 李润祥,高小红,汤敏. 基于双树复小波分解的Boosting集成学习土地覆被分类研究[J]. 遥感技术与应用, 2022, 37(2): 354-367.
[11] 李永康,王新军,马燕飞,陈蓓,闫立男,张冠宏. 基于机器学习算法的AMSR-2地表温度降尺度研究[J]. 遥感技术与应用, 2022, 37(2): 474-487.
[12] 刘倩,胡心雨,李晓彤,覃先林. 一种像素与对象相结合的林区建筑物识别方法[J]. 遥感技术与应用, 2021, 36(6): 1350-1357.
[13] 姜玉峰,齐建国,陈博伟,闫敏,黄龙吉,张丽. 基于无人机高光谱影像和机器学习的红树林树种精细分类[J]. 遥感技术与应用, 2021, 36(6): 1416-1424.
[14] 祝一诺,高婷,王术东,周磊,杜明义. 基于迁移学习再训练模型和高分遥感数据的建筑垃圾自动识别方法[J]. 遥感技术与应用, 2021, 36(2): 314-323.
[15] 徐红涛,陈春波,郑宏伟,罗格平,杨辽,王伟胜,吴世新. 集成建模变量优选和参数学习的SVR盐渍化监测[J]. 遥感技术与应用, 2021, 36(1): 176-186.