Please wait a minute...
img

官方微信

遥感技术与应用  2023, Vol. 38 Issue (1): 163-172    DOI: 10.11873/j.issn.1004-0323.2023.1.0163
数据与图像处理     
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
张妮娜1,2(),张珂1,2,3,4,5(),李运平1,2,李曦1,2,刘涛2,1
1.河海大学 水文水资源与水利工程科学国家重点实验室,江苏 南京 210098
2.河海大学 水文水资源学院,江苏 南京 210098
3.长江保护与绿色发展研究院,江苏 南京 210098
4.中国气象局-河海大学水文气象研究联合实验室,江苏 南京 210098
5.水利部水利大数据重点实验室,江苏 南京 210098
Study on Machine Learning Methods for Vegetation Classification in Typical Humid Mountainous Areas of South China based on the UAV Multispectral Remote Sensing
Nina ZHANG1,2(),Ke ZHANG1,2,3,4,5(),Yunping LI1,2,Xi LI1,2,Tao LIU2,1
1.State Key Laboratory of Hydrology-Water Recourses and Hydraulic Engineering,Hohai University,Nanjing 210098,China
2.College of Hydrology and Water recourses,Hohai University,Nanjing 210098,China
3.Yangtze Institute for Conservation and Development,Nanjing 210098,China
4.CMA-HHU Joint Laboratory for Hydrometeorological Studies,Nanjing 210098,China
5.Key Laboratory of Water Big Data Technology of Ministry of Water Resources, Hohai University, Nanjing 210098, China
 全文: PDF(3257 KB)   HTML
摘要:

为探究不同机器学习模型在我国南方典型湿润山区的植被类型分类效果,基于无人机遥感影像、实地观测数据、数字高程模型建立遥感特征,选用决策树、随机森林、支持向量机和AdaBoost模型在安徽屯溪流域构建植被类型遥感分类模型;选择总体精度、Kappa系数、均方误差、用户精度和生产者精度等评价指标,分析对比4种机器学习模型在典型小流域的适用性。结果表明:对于林地类型,AdaBoost模型分类精度最高,表明AdaBoost模型在林地分类中有明显的优势;对于非林地类型,模型之间精度差异较大,随机森林模型精度最高;整体而言,4种模型在南方典型湿润山区典型小流域均可获得较好的分类效果,其中AdaBoost模型总体精度为95.55%、Kappa系数为0.9419,均为最高,支持向量机模型表现均最低。地形因子、纹理特征等辅助特征为分类过程提供了重要信息,有助于提高分类精度。

关键词: 无人机遥感植被分类机器学习决策树随机森林支持向量机AdaBoost    
Abstract:

To explore the capabilities of a set of machine learning methods for vegetation classification in typical humid mountainous areas of south China, four types of the machine learning models, including Decision Tree (DT), Random Forest (RF), Support Vector Machine (SVM) and Adaptive Boosting (AdaBoost), were used to build the vegetation classification methods based on the UAV remote sensing images, field observation data, and digital elevation models. A suit of performance matrics such as classification accuracy, kappa coefficient, mean square error, user accuracy, and producer precision were selected to quantify the performance of the four machine learning methods. The results show that the AdaBoost model has the highest classification accuracy for identifying the forest vegetation types indicating that the AdaBoost model has an obvious advantage for distinguishing the forest vegetation types. Regarding the classification of non-forest types, the performances of the four methods differ relatively large with the highest accuracy in the RF model. In general, the four models can achieve good classification results in the typical humid mountainous areas of south China. The AdaBoost model has the highest classification accuracy and Kappa coefficient, while the SVM model has the relatively lowest accuracy. Auxiliary feature information such as topographic factors and texture features provide important information for improving the classification accuracy.

Key words: UAV remote sensing    Vegation classification    Machine learning    Decision tree    Random forest    Support vector machine    AdaBoost
收稿日期: 2021-10-27 出版日期: 2023-04-12
ZTFLH:  TP79  
基金资助: 国家自然科学基金项目(51879067);江苏省杰出青年基金项目(BK20180022);江苏省“六大人才高峰”高层次人才项目(NY-004)
通讯作者: 张珂     E-mail: ninazhang@hhu.edu.cn;kzhang@hhu.edu.cn
作者简介: 张妮娜(1997-),女,江苏苏州人,硕士研究生,主要从事遥感水文研究。E?mail: ninazhang@hhu.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  
张妮娜
张珂
李运平
李曦
刘涛

引用本文:

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.

Nina ZHANG,Ke ZHANG,Yunping LI,Xi LI,Tao LIU. Study on Machine Learning Methods for Vegetation Classification in Typical Humid Mountainous Areas of South China based on the UAV Multispectral Remote Sensing. Remote Sensing Technology and Application, 2023, 38(1): 163-172.

链接本文:

http://www.rsta.ac.cn/CN/10.11873/j.issn.1004-0323.2023.1.0163        http://www.rsta.ac.cn/CN/Y2023/V38/I1/163

图1  研究区地理位置、无人机遥感影像及样点示意图
特征类别参数个数
光谱特征ρNirρRedρGreen3
植被指数NDVI、RVI 、SAVI3
地形因子DEM、Slope、Aspect、日照时数4
纹理特征

MEA、VAR、HOM、CON、

DIS、ENT、SM、COR

8
表1  遥感特征选取列表
图2  单个机器学习模型下植被分类结果
类别DTRFSVMAdaBoost
UA/%PA/%UA/%PA/%UA/%PA/%UA/%PA/%
杉树80.3188.7087.3990.4383.0885.7190.3794.57
竹林90.8383.1990.3294.1279.6678.3397.5496.74
杉竹混合林94.5192.4796.6793.5578.9580.6597.7595.60
乔木混合林96.1097.3797.3396.0595.8392.0097.5098.73
灌木丛100.0076.47100.0064.7192.3185.71100.00100.00
农作物80.2694.7490.48100.0092.3192.31100.0076.92
裸地83.3371.43100.0085.7157.14100.00100.0066.67
水体100.00100.00100.00100.00100.0025.00100.00100.00
总体精度/%89.8792.4383.5295.55
Kappa系数0.861 60.903 70.787 60.941 9
均方误差0.305 10.405 30.514 50.376 4
表2  基于不同机器学习模型的分类精度评价
图3  DT、RF、AdaBoost模型中遥感特征贡献度
图4  不同组合方案下DT、RF、SVM、AdaBoost模型的总体精度、Kappa系数和均方误差
1 ZHANG K,ALI A, ANTONARAKIS A, et al. The sensitivity of North American Terrestrial carbon fluxes to spatial and temporal variation in soil moisture:An analysis using radar‐derived estimates of root‐aone soil moisture[J]. Journal of Geophysical Research:Biogeosciences,2019,124(11):3208-3231.
2 LIU Hongwei, GAO Fei, YU Zhongbo, et al. Study on temporal-spatial variability of soil moisture content on hillslope in a humid area[J]. Water Resources Protection,2016,32(5): 17-23.
2 刘宏伟, 高菲, 余钟波, 等. 湿润地区坡面土壤含水率时空变异性研究[J]. 水资源保护,2016,32(5):17-23.
3 LI Qiong, GONG Junfu, WANG Hongxin, et al. Application and comparison of hydrological models in small watersheds of humid regions[J]. South-to-North Transfers and Water Science & Technology, 2020, 18(4):81-86.
3 李琼, 龚珺夫, 王洪心, 等. 湿润地区山区小流域水文模型应用与比较[J]. 南水北调与水利科技(中英文), 2020, 18(4): 81-86.
4 LEE E, LIVINO A, HAN S, et al. Land cover change explains the increasing discharge of the Parana River[J]. Regional environmental change, 2018, 18(6): 1871-1881.
5 ZHANG K, KIMBALL J S, NEMANI R R, et al. Vegetation greening and climate change promote multidecadal rises of global land evapotranspiration[J]. Scientific Reports, 2015, 5(1): 1-9.
6 DU Yong, LI Jianzhu, NIU Kaijie, et al. Analysis of vegetation change and its impact on natural runoff in the mountain area of the Yongding River Basin from 1982 to 2015[J], Journal of Hydraulic Engineering,2021,52(11):1309-1323.
6 杜勇,李建柱,牛凯杰,等. 1982~2015年永定河山区植被变化及对天然径流的影响[J]. 水利学报,2021,52(11):1309-1323.
7 QIAO L, ZOU C B, STEBLER E, et al. Woody plant encroachment reduces annual runoff and shifts runoff mechanisms in the tallgrass prairie, USA[J]. Water Resources Research, 2017, 53(6): 4838-4849.
8 LIU He, GU Lingjia, REN Ruizhi. Research progress of forest parameter acquisition based on UAV remote sensing technology[J]. Remote Sensing Technology and Application, 2021, 36(3): 489-501.
8 刘鹤, 顾玲嘉, 任瑞治. 基于无人机遥感技术的森林参数获取研究进展[J]. 遥感技术与应用, 2021, 36(3): 489-501.
9 ZHANG Ke, LIU Linxin, CHAO Lijun, et al. Spatiotemporal variations of terrestrial ecosystem water use efficiency in Yunnan Province from 2000 to 2014[J], Water Resources Protection,2019,35(5):1-5.
9 张珂, 刘林鑫, 晁丽君, 等. 2000~2014年云南省陆地生态系统水分利用效率的时空变化[J]. 水资源保护, 2019, 35(5):1-5.
10 FU Wei, XU Hanqiu, WANG Meiya, et al. Vegetation classification and variation assessment in a typical red soil erosion area in Southern China:Hetian, Changting Country of Fujian Province[J]. Remote Sensing Technology and Application, 2017, 32(3): 546-555.
10 付伟, 徐涵秋, 王美雅, 等. 南方红壤典型水土流失区植被分类及植被类型变化的遥感评估——以福建省长汀县河田地区为例[J]. 遥感技术与应用, 2017, 32(3): 546-555.
11 YANG Hongyan, DU Jianmin, RUAN Peiying, et al. Vegetation classification of desert steppe based on unmanned aerial vehocle remote sensing and random forest[J]. Transactions of the Chinese Society for Agricultural Machinery,2021,52(6): 186-194.
11 杨红艳, 杜健民, 阮培英, 等. 基于无人机遥感与随机森林的荒漠草原植被分类方法[J]. 农业机械学报, 2021,52(6):186-194.
12 COLOMINA I, MOLINA P. Unmanned aerial systems for photogrammetry and remote sensing: A review - ScienceDirect[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2014, 92(2): 79-97.
13 HU Yina, AN Ru, AI Zetian, et al. Researches on grass species fine identification based on UAV hyperspectral images in three-river source region[J]. Remote Sensing Technology and Application,2021,36(4):926-935.
13 胡宜娜, 安如, 艾泽天, 等. 基于无人机高光谱影像的三江源草种精细识别研究[J]. 遥感技术与应用,2021,36(4):926-935.
14 LI Hongda, GAO Xiaohong, TANG Min. Land cover classification for different spatial resolution images from CNN[J]. Remote Sensing Technology and Application, 2020, 35(4): 749-758.
14 李宏达, 高小红, 汤敏. 基于CNN的不同空间分辨率影像土地覆被分类研究[J]. 遥感技术与应用, 2020, 35(4): 749-758.
15 WANG Hongshu, LI Shihua, GUO Jiawei, et al. Retrieval of the leaf area density of Magnolia woody canopy with terrestrial Laser-scanning data[J]. Journal of Remote Sensing, 2016, 20(4): 570–578.
15 王洪蜀, 李世华, 郭加伟, 等. 地基激光雷达的玉兰林冠层叶面积密度反演[J]. 遥感学报, 2016, 20(4): 570-578.
16 LI Nan. Long-term remote sensing dynamic monitoring and ecological assessment of the coastal wetland in Hangzhou Bay[D]. Nanjing: Nanjing Forestry University, 2020.
16 李楠. 杭州湾滨海湿地长时间尺度遥感动态监测及生态评估[D]. 南京: 南京林业大学, 2020.
17 YAO Lianqiu. Research on remote sensing image land cover classification based on expert knowledge and improved SEATH algorithm[D]. Nanjing: Nanjing Normal University, 2018.
17 姚恋秋. 基于专家知识和改进SEATH算法的遥感影像土地覆被分类研究[D]. 南京: 南京师范大学, 2018.
18 SONG Minghui. Object-oriented urban land classfication with GF-2 remote sensing image[J]. Remote Sensing Technology and Application, 2019, 34(3): 547-552.
18 宋明辉. 基于高分二号数据的面向对象城市土地利用分类研究[J]. 遥感技术与应用, 2019, 34(3): 547-552.
19 GILBERTSON J K, KEMP J, NIEKERK A V. Effect of pan-sharpening multi-temporal Landsat 8 imagery for crop type differentiation using different classification techniques[J]. Computers & Electronics in Agriculture,2017,134(5):151-159.
20 RACZKO E, ZAGAJEWSKI B.Comparison of support vector machine, random forest and neural network classifiers for tree species classification on airborne hyperspectral APEX images[J].European Journal of Remote Sensing,2017,50(1):144-154.
21 KUMAR P, GUPTA D K, MISHRA V N, et al. Comparison of support vector machine, artificial neural network, and spectral angle mapper algorithms for crop classification using LISS IV data[J]. International Journal of Remote Sensing, 2015, 36(5-6): 1604-1617.
22 GILBERTSON J K. Machine learning for object-based crop classification using multi-temporal Landsat 8 imagery[D]. Stellenbosch: Stellenbosch University, 2017.
23 SHAO Wenjing, SUN Weiwei, YANG Gang. Comparison of texture feature extraction methods for hyperspectral imagery classification[J]. Remote Sensing Technology and Application, 2021, 36(2): 431-440.
23 邵文静, 孙伟伟, 杨刚. 高光谱遥感影像纹理特征提取的对比分析[J]. 遥感技术与应用, 2021, 36(2): 431-440.
24 WANG Huaijing, TAN Bingxiang, FANG Xiufeng, et al. Precise classification of forest types use Hyperion image based on the C5.0 decision tree algorithm[J]. Journal of Zhejiang A&F University, 2018, 35(4): 147-157.
24 王怀警, 谭炳香, 房秀凤, 等. C5.0决策树Hyperion影像森林类型精细分类方法[J]. 浙江农林大学学报, 2018, 35(4): 147-157.
25 Technical regulations for inventory for forest management planning and design [S]. 2011,GB/T 26424-2010:36.
25 森林资源规划设计调查技术规程 [S].2011,GB/T 26424-2010:36.
26 HUETE A, DIDAN K, MIURA T, et al. Overview of the radiometric and biophysical performance of the MODIS vegetation indices[J]. Remote Sensing of Environment, 2002, 83(1-2): 195-213.
27 LI Xiang, LIU Kai, ZHU Yuanhui, et al. Study on the mangrove species classification based on ZY-3 image[J]. Remote Sensing Technology and Application, 2018,33(2):360-369.
27 李想, 刘凯, 朱远辉, 等. 基于资源三号影像的红树林物种分类研究[J]. 遥感技术与应用, 2018,33(2):360-369.
28 ZHANG Yu, ZHANG Liyuan, ZHANG Huihui, et al. Crop coefficient estimation method ofmaize by UAV remote sensing and soil moisture monitoring[J]. Transactions of the Chinese Society of Agricultural Engineering,2019,35(1):83-89.
28 张瑜, 张立元, Zhang Huihui, 等. 玉米作物系数无人机遥感协同地面水分监测估算方法研究[J]. 农业工程学报, 2019, 35(1): 83-89.
29 HUETE A R. A Soil-Adjusted Vegetation Index(SAVI)[J]. Remote Sensing of Environment, 1988, 25(3): 295-309.
30 FENG Jianhui, YANG Yujing. Research on extracting texture feature image based on gray level co-occurrence matrix[J]. Beijing Surveying and Mapping, 2007, 4(3): 19-22.
30 冯建辉, 杨玉静. 基于灰度共生矩阵提取纹理特征图像的研究[J]. 北京测绘, 2007, 4(3): 19-22.
31 FRIEDL M A, BRODLEY C E. Decision tree classification of land cover from remotely sensed data[J]. Remote Sensing of Environment, 1997, 61(3): 399-409.
32 BREIMAN L. Random forest[J]. Machine Learning, 2001, 45(1): 5-32.
33 VAPNIK V.Statistical Learning Theory[M]. New York:Wiley,1998.
34 FREUND Y, SCHAPIRE R E. A decision-theoretic generali-zation of on-line learning and an application to boosting[J]. Journal of Computer and System Sciences,1997,55(1):119-139.
[1] 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.
[2] 范霄,孔金玲,钟炎伶,蒋镒竹,张静雅. 基于XGBoost算法的遥感图像云检测[J]. 遥感技术与应用, 2023, 38(1): 156-162.
[3] 马超,黄华国,田昕,刘炳杰,温坤剑,王鹏杰. 基于背包激光雷达的单木材积估测方法研究[J]. 遥感技术与应用, 2022, 37(5): 1071-1083.
[4] 谢雨涵,史建康,孙晓慧,吴文瑾,李新武. 基于大疆精灵4无人机多光谱影像的西沙植被监测[J]. 遥感技术与应用, 2022, 37(5): 1170-1178.
[5] 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.
[6] 张琪,张光辉,张岩,王佳希,余双武. 基于不同分辨率遥感影像自动提取切沟的精度分析和转换模型[J]. 遥感技术与应用, 2022, 37(5): 1217-1226.
[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] 陈绎冰,李天依,李欣艳,赵文川,范荣峰,陈凤娇,杨元建. 基于随机森林和遥感的台风降水云光谱与降水关系研究[J]. 遥感技术与应用, 2022, 37(5): 1277-1288.
[9] 丁书培,李蒙蒙,汪小钦,李琳,吴瑞姣,黄姮. 基于时序遥感数据的福州市耕地非农化特征及驱动因子分析[J]. 遥感技术与应用, 2022, 37(3): 550-563.
[10] 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.
[11] 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.
[12] 李媛,周小成,陈芸芝,王锋克. 结合Landsat时序遥感干扰检测的福建将乐县森林林龄估算与评价[J]. 遥感技术与应用, 2022, 37(3): 651-662.
[13] 王美雅,徐涵秋. 超大城市土地覆盖与热环境的随机森林回归模型研究[J]. 遥感技术与应用, 2022, 37(2): 379-388.
[14] 康新礼,张文豪,刘原萍,顾行发,余涛,张丽丽,徐桦昆. 基于随机森林的京津冀地区PM2.5遥感反演及变化分析[J]. 遥感技术与应用, 2022, 37(2): 424-435.
[15] 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.