Please wait a minute...
img

官方微信

遥感技术与应用  2022, Vol. 37 Issue (2): 368-378    DOI: 10.11873/j.issn.1004-0323.2022.2.0368
LUCC专栏     
基于CNN的吉林一号卫星城市土地覆被制图潜力评估
吕冬梅1(),马玥2,3(),李华朋3
1.吉林建筑大学 电气与计算机学院,吉林 长春 130118
2.吉林建筑大学 测绘与勘查工程学院,吉林 长春 130118
3.中国科学院 东北地理与农业生态研究所,吉林 长春 130102
Evaluating the Potential of JL1 Remote Sensing Data in Urban Land Cover Classification Using Convolutional Neural Networks
Lü Dongmei1(),Yue Ma2,3(),Huapeng Li3
1.School of Electrical and Computer Engineering,Jilin Jianzhu University,Changchun 130118,China
2.School of Geomatics and Prospecting Engineering,Jilin Jianzhu University,Changchun 130118,China
3.Northeast Institute of Geography and Agroecology,CAS,Changchun 130102,China
 全文: PDF(7791 KB)   HTML
摘要:

以吉林一号视频07B星高分遥感影像为基础,采用卷积神经网络(CNN)对城区土地覆被进行精细分类,设置多组光谱变量集合,并与最大似然法、多层感知机和支持向量机分类方法进行对比,全面评估分析各方法对城区土地覆被信息提取的适用性及波谱特征对分类精度的影响。结果表明:CNN模型的分类精度最高,总体精度高于90%,相比其他方法提高幅度达12%以上,能够显著降低“椒盐”噪音;红边波段对所有方法总体分类精度贡献十分有限,而近红外波段对分类精度的提升较为明显;总体而言,红边和近红外波段对CNN分类精度影响较为微弱。深度学习应用于吉林一号高分遥感数据能获取高精度城区土地覆被分类图,可为城市土地资源配置,城市规划与管理提供重要的支撑。

关键词: 吉林一号卷积神经网络土地覆被分类深度学习高分影像    
Abstract:

This research classified urban land cover using the Convolutional Neural Network (CNN) model based on the fine spatial resolution remotely sensed imagery from recently launched JL1 07B satellite. We applied CNN to classify imagery using different combinations of spectral feature variables, and compared the performance of CNN with three other methods, namely maximum likelihood classification algorithm, multi-layer perceptron algorithm and support vector machine algorithm. The experimental results demonstrated that CNN consistently achieved the highest overall accuracy (>90%), larger than that of other methods by above 12%, and reduced significantly the “salt-and-pepper” noise. The contribution of red-edge band to the classification accuracy was slight, while the near-infrared (NIR) band could increase the OA prominently. Overall, the effect of red-edge and near-infrared bands exerted a slightly impact on the OA of CNN, demonstrating the robustness and generalization of the CNN model. The high accuracy urban land cover classification map achieved using CNN based on JL1 satellite imagery can support the decision makings for land resource allocation, urban planning and regional administration.

Key words: JL1 satellite    Convolutional neural network    Land cover    Deep learning    Fine spatial resolution imagery
收稿日期: 2021-01-13 出版日期: 2022-06-17
ZTFLH:  P237  
基金资助: 国家自然科学基金项目(41730104);吉林省教育厅“十三五”科学技术项目(JJKH20180595KJ)
通讯作者: 马玥     E-mail: dongmeilv@126.com;mayue417@hotmail.com
作者简介: 吕冬梅(1975-),女,吉林长春人,副教授,主要从事人工智能与图像处理研究。E?mail:dongmeilv@126.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  
吕冬梅
马玥
李华朋

引用本文:

吕冬梅,马玥,李华朋. 基于CNN的吉林一号卫星城市土地覆被制图潜力评估[J]. 遥感技术与应用, 2022, 37(2): 368-378.

Lü Dongmei,Yue Ma,Huapeng Li. Evaluating the Potential of JL1 Remote Sensing Data in Urban Land Cover Classification Using Convolutional Neural Networks. Remote Sensing Technology and Application, 2022, 37(2): 368-378.

链接本文:

http://www.rsta.ac.cn/CN/10.11873/j.issn.1004-0323.2022.2.0368        http://www.rsta.ac.cn/CN/Y2022/V37/I2/368

图1  研究区及重点实验区范围示意图
图2  卷积神经网络结构示意图
S1实验区S2实验区
合计3 5363 5063 6033 521
土地覆被类型类型代码训练样本/个测试样本/个类型代码训练样本/个测试样本/个
混凝土屋顶C1545541C1505507
金属屋顶C2508501C2418352
黏土屋顶C3358371C3461485
塑胶表面C4314309C4310293
沥青路面C5440438C5481483
铁路C6226221C7510485
林地C7516513C8314311
草地C8317303C9300301
裸土C9312309C10304304
表1  样本统计表
模型S1实验区S2实验区
OA/%Kappa系数OA/%Kappa系数
MLC73.050.7078.480.76
MLP74.190.7179.420.77
SVM72.990.6978.650.76
CNN86.570.8591.470.90
表2  不同分类方法总体精度和Kappa系数对比
地物类型S1实验区 PA/%地物类型S2实验区 PA/%
MLCMLPSVMCNNMLCMLPSVMCNN
混凝土屋顶32.3556.9349.1786.69混凝土屋顶46.7557.4067.8589.55
金属屋顶100.0099.0099.0097.41金属屋顶95.4594.8994.0395.45
黏土屋顶70.0874.6678.4488.41黏土屋顶81.4484.3387.8496.91
塑胶表面60.5257.6155.3477.99塑胶表面77.1352.2236.1887.71
沥青路面77.6370.3262.3376.48沥青路面75.7876.4070.3979.50
铁路78.7357.0172.4087.78林地97.3898.4797.6096.94
林地96.8899.4298.8396.10草地88.7587.7888.1094.86
草地77.5678.8875.2586.80裸土67.7773.4266.1191.03
裸土62.1451.4654.0572.49水体84.2190.1392.4392.76
表3  不同分类方法生产者精度对比
S1实验区混凝土屋顶金属屋顶黏土屋顶塑胶表面沥青路面铁路林地草地裸土总和UA/%
混凝土屋顶4691218053122001359778.56
金属屋顶04880000000488100
黏土屋顶12032830000034395.63
塑胶表面001241170002528484.86
沥青路面261013352012739385.24
铁路16012019194001425576.08
林地00003249339654390.79
草地00023200263028891.32
裸土18012419110022431571.11
总和5415013713094382215133033093 506
PA/%86.6997.4188.4177.9976.4887.7896.186.872.49
OA/%86.57
Kappa0.847 4
表4  S1实验区CNN模型的混淆矩阵
S2实验区混凝土屋顶金属屋顶黏土屋顶塑胶表面沥青路面林地草地裸土水体总数UA/%
混凝土屋顶454165919000050390.26
金属屋顶03360017000035395.18
黏土屋顶40470130000048796.51
塑胶表面80102572000027792.78
沥青路面170012384010041492.75
林地000016444902249190.43
草地00001329527032690.49
裸土15002606274030390.43
水体900038110028234082.94
总数5073524852934834583113013043 494
PA/%89.5595.4596.9187.7179.5096.9494.8691.0392.76
OA/%91.47
Kappa0.903 5
表5  S2实验区CNN模型的混淆矩阵
图3  不同波谱组合的总体精度和Kappa系数对比图
图4  不同地物类型生产者精度变化率
图5  不同分类方法城区土地覆被分类对比图
1 Zhang Xueru, Yao Yifeng, Kong Shaojun, et al. The analysis on urban costruction land change and driving forces in Nanjing from 2000 to 2014[J]. Resources and Environment in the Yangtze Basin,2017,26(4): 552-562.
1 张雪茹, 姚亦锋, 孔少君,等. 南京市2000~2014年城市建设用地变化及驱动因子研究[J]. 长江流域资源与环境, 2017,26(4):552-562.
2 Zhang Chao, Li Zhixiao, Li Pengshan, et al. Urban-rural land use plan monitoring based on high spatial resolution remote sensing imagery classification[J].Transactions of the Chinese Society of Agricultural Machinery, 2015, 46(11): 323-329.
2 张超, 李智晓, 李鹏山, 等.基于高分辨率遥感影像分类的城镇土地利用规划监测[J].农业机械学报,2015,46(11):323-329.
3 Fang Tao, Wang Dongzhi. Land use change analysis in the coastal aone of Zhuhai city based on ULAT method[J]. Geospatial Information, 2020, 18(3): 69-71.
3 方涛, 王冬至. 基于ULAT的珠海市海岸带土地利用变化分析[J]. 地理空间信息, 2020, 18(3): 69-71.
4 Lo C P, Choi J.A hybrid approach to urban land use/cover map-ping using Landsat 7 enhanced thematic Mapper Plus (ETM+) images[J]. International Journal of Remote Sensing,2004,25(14):2687-2700. DOI:10.1080/0143116031000 1618428 .
doi: 10.1080/0143116031000 1618428
5 Hu T Y, Yang J, Li X, et al. Mapping urban land use by using landsat images and open social data[J]. Remote Sensing, 2016, 8(2): 151. DOI: 10.3390/rs8020151 .
doi: 10.3390/rs8020151
6 Sun Danfeng, Yang Jihong, Liu Shunxi. Application of high-spatial IKNOS remote sensing images in land use classification and change monitoring[J]. Transactions of the Chinese Society of Agricultural Engineering, 2002,18(2): 160-164.
6 孙丹峰, 杨冀红, 刘顺喜. 高分辨率遥感卫星影像在土地利用分类及其变化监测的应用研究[J]. 农业工程学报,2002,18(2):160-164.
7 Mawenda J, Watanabe T, Avtar R. An analysis of urban land use/land cover changes in blantyre city, Southern Malawi (1994-2018)[J]. Sustainability,2020,12(6):2377. DOI:10. 3390/su12062377 .
doi: 10. 3390/su12062377
8 Feng Xin, Du Shihong, Zhang Fangli, et al. Urban land classification of high-resolution images based on multi-scale fusion[J]. Geography and Geo-Information Science, 2013, 29(3): 43-47.
8 冯昕, 杜世宏, 张方利, 等. 基于多尺度融合的高分辨率影像城市用地分类[J]. 地理与地理信息科学, 2013, 29(3): 43-47.
9 Herold M, Liu X H, Clarke K C. Spatial metrics and image texture for mapping urban land use[J]. Photogrammetric Engineering and Remote Sensing, 2003, 69(9):991-1001. DOI: 10.14358/PERS.69.9.991 .
doi: 10.14358/PERS.69.9.991
10 Ziaei Z, Pradhan B, Mansor S, et al. A rule-based parameter aided with object-based classification approach for extraction of building and roads from WorldView-2 images[J]. Geocarto International,2014,29(5):554-569. DOI:10.1080/10106049. 2013.819039 .
doi: 10.1080/10106049. 2013.819039
11 Zhang Chenyang, Yang Changkun, Jia Mengna, et al. Change detection of land use in Wuhan City based on multi temporal band combination of GF-1[J]. Geomatics & Spatial Information Technology, 2016, 39(9): 33-35.
11 张晨阳,杨长坤, 贾萌娜, 等. 基于GF-1多时相波段组合的武汉市土地利用变化检测[J].测绘与空间地理信息, 2016, 39(9): 33-35.
12 Zhu Jishuai, Deng Guang, Tian Deyu, et al. Application of JL1 high-resolution image in urban management[J]. Satellite Application, 2020(3): 54-60.
12 朱济帅, 邓广, 田德宇, 等. 吉林一号高分辨率影像在城市管理中的应用[J]. 卫星应用, 2020(3): 54-60.
13 Han Rui, Huang Peng, Lu Nan, et al. Application of JL1 in surface water remote sensing monitoring[J]. Satellite Application,2020,99(3):49-55.
13 韩睿,黄鹏,芦楠,等.吉林一号在地表水遥感监测中的应用[J].卫星应用,2020,99(3):49-55.
14 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 Technology and Application, 2019, 34(1): 57-67.
14 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):57-67.
15 Deng Zeng, Li Dan, Ke Yinghai, et al. An Improved SVM algorithm for high spatial resolution remote sensing image classification[J]. Remote Sensing for Land and Resources, 2016, 28(3): 12-18.
15 邓曾, 李丹, 柯樱海, 等. 基于改进SVM算法的高分辨率遥感影像分类[J]. 国土资源遥感, 2016, 28(3): 12-18.
16 Yuan H, Van Der Wiele C F, Khorram S. An automated artificial neural network system for land use/land cover classification from Landsat TM imagery[J]. Remote Sensing. 2009, 1(3):243-265. DOI: 10.3390/rs1030243 .
doi: 10.3390/rs1030243
17 Marmanis D, Datcu M, Esch T, et al. Deep learning earth observation classification using imagenet pretrained networks[J]. IEEE Geoscience and Remote Sensing Letters, 2016, 13(1): 105-109. DOI: 10.1109/LGRS.2015.2499239 .
doi: 10.1109/LGRS.2015.2499239
18 Persello C, Stein A. Deep fully convolutional networks for the detection of informal settlements in VHR images[J]. IEEE Geoscience and Remote Sensing Letters, 2017, 14(12): 2325-2329. DOI: 10.1109/LGRS.2017.2763738 .
doi: 10.1109/LGRS.2017.2763738
19 Maggiori E, Tarabalka Y, Charpiat G, et al. High-resolution aerial image labeling with convolutional neural networks[J]. IEEE Transactions on Geoscience and Remote Sensing, 2017,55(12):7092-7103. DOI:10.1109/TGRS.2017. 2740362 .
doi: 10.1109/TGRS.2017. 2740362
20 Kussul N, Lavreniuk M, Skakun S, et al. Deep learning classification of land cover and crop types using remote sensing data[J]. IEEE Geoscience and Remote Sensing Letters, 2017, 14(5): 778-782. DOI: 10.1109/LGRS.2017.2681128 .
doi: 10.1109/LGRS.2017.2681128
21 Mei S H, Ji J Y, Hou J H, et al. Learning sensor-specific spatial-spectral features of hyperspectral images via convolutional neural networks[J]. IEEE Transactions on Geoscience and Remote Sensing, 2017, 55(8):4520-4533.
22 Tao Y T, Xu M Z, Zhang F, et al. Unsupervised-restricted deconvolutional neural network for very high resolution remote-sensing image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2017, 55(12): 6805-6823. DOI: 10.1109/TGRS.2017.2734697 .
doi: 10.1109/TGRS.2017.2734697
23 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.
23 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.
24 Zhou Zhuang, Li Shengyang, Zhang Kang, et al. Crop mapping using remotely sensed spectral and context features based on CNN[J]. Remote Sensing Technology and Application, 2019, 34(4): 694-703 .
24 周壮, 李盛阳, 张康, 等. 基于CNN和农作物光谱纹理特征进行作物分布制图[J]. 遥感技术与应用, 2019, 34(4): 694-703.
25 Li H P, Zhang C, Zhang S Q, et al. A hybrid OSVM-OCNN method for crop classification from fine spatial resolution remotely sensed imagery[J]. Remote Sensing, 2019, 11(20):2370. DOI: 10.3390/rs11202370 .
doi: 10.3390/rs11202370
26 Zhang C, Sargent I, Xin P, 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
27 Chen Leishi, Zhao Junsan, Li Yi, et al. On land use classification by means of machine learning based on multi-source remote sensing image fusion[J]. Journal of Southwest China Normal University(Natural Science Edition),2018,43(10):103-111.
27 陈磊士, 赵俊三, 李易, 等. 基于机器学习的多源遥感影像融合土地利用分类研究[J]. 西南师范大学学报: 自然科学版, 2018, 43(10): 103-111.
28 Li Quan, Wang Haiyan, Li Lin. A study of accuracy control of land cover classification based on MLH algorithm[J]. Scientific and Technological Management of Land and Resources, 2005, 22(4): 42-45.
28 李全, 王海燕, 李霖. 基于最大似然分类算法的土地覆盖分类精度控制研究[J]. 国土资源科技管理, 2005, 22(4): 42-45.
29 Han Ling. The classification model of RS images based on artificial neural network——MLP[J]. Bulletin of Surveying and Mapping, 2004(9): 29-30.
29 韩玲. 基于人工神经网络——多层感知器(MLP)的遥感影像分类模型[J]. 测绘通报, 2004(9): 29-30.
30 Wu Xueming, Zhang Huaiqing, Lin Hui,et al. Wetland change forecast in Beijing based on support vector machines[J]. Chinese Agricultural Science Bulletin, 2012, 28(14): 280-284.
30 吴学明, 张怀清, 林辉, 等. 基于支持向量机的北京市湿地变化预测研究[J]. 中国农学通报, 2012, 28(14): 280-284.
[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] 杨雪峰. 使用高分遥感影像获取塔里木河胡杨高度信息[J]. 遥感技术与应用, 2021, 36(5): 1199-1208.
[3] 郑磊,何直蒙,丁海勇. 基于ENVINet5的高分辨率遥感影像稀疏塑料大棚提取研究[J]. 遥感技术与应用, 2021, 36(4): 908-915.
[4] 路春燕,雷依凡,苏颖,黄雨菲,刘明月,贾明明. 基于面向对象—深度学习的闽东南低海拔海岸带地区湿地动态遥感分析[J]. 遥感技术与应用, 2021, 36(4): 713-727.
[5] 刘钊,赵桐,廖斐凡. 基于卷积神经网络预测结果的缝隙修复算法研究[J]. 遥感技术与应用, 2021, 36(2): 247-255.
[6] 屈炀,袁占良,赵文智,陈学泓,陈家阁. 基于多时序特征和卷积神经网络的农作物分类[J]. 遥感技术与应用, 2021, 36(2): 304-313.
[7] 李庆,陈俊杰,李庆亭,李柏鹏,卢凯旋,昝露洋,陈正超. 基于SSD模型的京津冀地区尾矿库检测[J]. 遥感技术与应用, 2021, 36(2): 293-303.
[8] 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.
[9] 陈敏,潘佳威,李江杰,徐璐,刘加敏,韩健,陈奕云. 结合VGGNet与Mask R-CNN的高分辨率遥感影像建设用地检测[J]. 遥感技术与应用, 2021, 36(2): 256-264.
[10] 林娜,冯丽蓉,张小青. 基于优化Faster-RCNN的遥感影像飞机检测[J]. 遥感技术与应用, 2021, 36(2): 275-284.
[11] 胡乃勋,陈涛,甄娜,牛瑞卿. 基于卷积神经网络的面向对象露天采场提取[J]. 遥感技术与应用, 2021, 36(2): 265-274.
[12] 徐瑾昊,冯敏,王建邦,冉有华,祁元,杨联安,李新. 基于高分遥感数据和深度学习的石冰川自动提取研究[J]. 遥感技术与应用, 2020, 35(6): 1329-1336.
[13] 张永宏,曹海啸,阚希. 基于FY-4A/AGRI时空特征融合的新疆地区积雪判识[J]. 遥感技术与应用, 2020, 35(6): 1337-1347.
[14] 杨瑞,祁元,苏阳. 深度学习U-Net方法及其在高分辨卫星影像分类中的应用[J]. 遥感技术与应用, 2020, 35(4): 767-774.
[15] 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.