Please wait a minute...
img

官方微信

遥感技术与应用  2021, Vol. 36 Issue (3): 705-712    DOI: 10.11873/j.issn.1004-0323.2021.3.0705
遥感应用     
基于光谱聚类的高分影像复杂地块特征提取方法研究
闫润州1,2(),李利伟2(),王涛3,陈俊奇3,赖健3,张兵1,2
1.中国科学院大学,北京 100049
2.中国科学院空天信息创新研究院 中国科学院数字地球重点实验室 北京 100094
3.上海卫星工程研究所 高分上海数据与应用中心 上海 201109
Spectral Clustering based Feature Extraction for Parcel Classification Using High Spatial Resolution Remote Sensing Images
Runzhou Yan1,2(),Liwei Li2(),Tao Wang3,Junqi Chen3,Jian Lai3,Bing Zhang1,2
1.University of Chinese Academy of Sciences,Beijing 100049,China
2.Key Laboratory of Digital Earth Science,Aerospace Information Research Institute,Chinese Academy of Sciences,Beijing 100094,China
3.Shanghai Data & Application Center of High-resolution Earth Observation System,Shanghai Institute of Satellite Engineering,Shanghai 201149,China
 全文: PDF(2365 KB)   HTML
摘要:

高空间分辨率遥感技术为大范围判识农用地利用类型提供了丰富的数据源。农用地类型多样性和复杂性给高效应用高分影像识别农用地类型带来很大挑战。地块矢量的引入可以帮助更好综合利用多元影像特征,进而提高农用地类型判识精度。但是,传统地块特征提取方法将地块视为一个整体,通过对地块内部像元特征平均实现地块特征表达,不能很好适用于地块内部像元光谱具有较强异质性的情况。针对内部光谱异质但具有较强规律的地块,提出一种基于光谱聚类的特征提取方法,将地块内部的光谱聚类特征作为地块的特征之一,对地块进行分类。利用上海崇明区内2个典型样区的BJ-2卫星影像和地面调查数据进行实验验证分析,结果表明:①该方法相对利用地块内部所有像元光谱平均的方法,能够有效提升地块分类精度;②通过引入地块内部光谱聚类特征到传统地块特征组合,可以进一步提升地块分类精度,对菜地和廊道等内部像元光谱混合比例变化较大的类别提升效果最为明显。该方法为复杂地块分类提供了新思路。

关键词: 北京2号地块特征聚类    
Abstract:

High spatial resolution satellite remote sensing provides redundant data for large-scale agricultural land management. Due to the variety and complexity of land parcels, it is still a great challenge to accurately and timely map land use types. Introducing parcel boundary has proven an effective strategy to integrate spatial and spectral features, to improve the classification accuracy. However, current feature extraction methods always treat each parcel as a whole and use only mean feature values of all pixels inside each parcel. This approach cannot well adapt to the scenarios that parcels include more than one types of spectral similar target. To this end, this paper proposes a spectral clustering based feature extraction method to better model the complexity of parcels. BJ-2 images and ground surveying data from 2 typical areas in the Chongming County in Shanghai were selected to experimentally evaluate the proposed method. The results show that: (1) Compared with the direct spectral averaging method, the proposed method can effectively improve the accuracy of land parcel classification; (2) Introducing the clustering features into the typical feature combination can further improve the accuracy of land parcel classification. And the improvement mainly lies on categories with unstable mixing ratio of internal pixel spectrum, such as vegetable field and corridor. The proposed method provides an effective alternative to classify parcel types especially for parcels including more than one spectral similar target.

Key words: BJ-2    Land parcels    Feature    Clustering
收稿日期: 2020-01-14 出版日期: 2021-07-22
ZTFLH:  TP753  
基金资助: 国家自然科学基金项目“中高分辨率多时相多源光学影像分类模型与方法研究”(41971327);国家重点研发计划项目“地球资源环境动态监测技术”(2016YFB0501501)
通讯作者: 李利伟     E-mail: azhouazhou5@163.com;lilw@radi.ac.cn
作者简介: 闫润州(1995—),男,浙江宁波人,硕士研究生,主要从事高分遥感图像与机器学习、深度学习结合方面的研究。E?mail:azhouazhou5@163.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  
闫润州
李利伟
王涛
陈俊奇
赖健
张兵

引用本文:

闫润州,李利伟,王涛,陈俊奇,赖健,张兵. 基于光谱聚类的高分影像复杂地块特征提取方法研究[J]. 遥感技术与应用, 2021, 36(3): 705-712.

Runzhou Yan,Liwei Li,Tao Wang,Junqi Chen,Jian Lai,Bing Zhang. Spectral Clustering based Feature Extraction for Parcel Classification Using High Spatial Resolution Remote Sensing Images. Remote Sensing Technology and Application, 2021, 36(3): 705-712.

链接本文:

http://www.rsta.ac.cn/CN/10.11873/j.issn.1004-0323.2021.3.0705        http://www.rsta.ac.cn/CN/Y2021/V36/I3/705

图1  光谱异质性地块
类别地块真彩色图像地块聚类结果光谱曲线示意
水稻
林地
廊道
菜地
水产地
大棚
表1  典型地块地物聚类结果示意
图2  光谱聚类特征提取流程
图3  研究区地理位置、遥感影像与地块真值图分布审图号:GS(2019)1822
编号类别名称样区1地块数量样区2地块数量
1水稻3222257
2林地2487852
3廊道-129
4菜地287133
5水产地733391
6大棚-33
表2  地块类别体系描述
特征名称特征说明
光谱特征地块内像元逐波段辐亮度平均值,实验采用蓝、绿、红、近红外4个波段
指数特征地块内像元逐个光谱指数平均值,实验采用指数包括NDVI、NDWI
纹理特征地块内像元逐波段纹理特征平均值,实验采用基于GLCM的熵和对比度(窗口为5,步长为1)
聚类特征本文提出的聚类特征(聚类个数为2)
表3  实验采用的地块特征说明
实验特征组合1特征组合2
1光谱特征聚类特征
2光谱特征光谱、聚类特征
3光谱、纹理特征光谱、纹理、聚类特征
4光谱、指数特征光谱、指数、聚类特征
5光谱、纹理、指数特征光谱、纹理、指数、聚类特征
表4  地块分类实验设计
样区实验

组合1

总体精度/%

组合2

总体精度/%

组合1

Kappa

组合2

Kappa

1171.6276.080.563 40.628 5
271.6276.730.563 40.636 8
381.1382.260.704 00.720 8
473.0576.640.582 00.636 0
581.8183.030.713 60.732 5
2178.4082.180.699 80.745 1
278.4082.520.699 80.752 1
380.1884.080.719 50.772 0
479.0682.960.704 90.754 6
582.7484.970.751 50.782 5
表5  实验总体精度
精度类别

地物

类别

实验1实验2实验3实验4实验5
组合1组合2组合1组合2组合1组合2组合1组合2组合1组合2
制图精度水稻69.1675.8069.1676.2382.2583.6673.1675.8083.1183.78
林地71.7973.5971.7975.3979.2380.2170.3275.1479.8982.01
菜地53.7459.8653.7456.4656.4657.1451.7059.8655.7856.46
水产地89.1092.1089.1091.5592.3792.9290.1992.1092.9293.73
用户精度水稻83.0485.1683.0485.7688.5089.0883.2985.4088.3288.57
林地68.9774.2668.9773.7681.0982.0269.7573.9381.8382.05
菜地24.0928.5724.0929.7529.7531.7023.3830.4531.6634.02
水产地80.1585.3580.1586.3889.6890.9387.8087.1189.7489.82
表6  实验数据1中类别用户精度和制图精度 (%)
精度类别

地物

类别

实验1实验2实验3实验4实验5
组合1组合2组合1组合2组合1组合2组合1组合2组合1组合2
制图精度水稻82.8186.7282.8185.9479.6984.3880.4783.5982.0382.03
林地77.0685.3277.0683.2681.6586.2479.3686.0186.2488.99
廊道75.0071.6775.0078.3361.6776.6771.6776.6766.6776.67
菜地42.6544.1242.6542.6550.0051.4738.2439.7144.1247.06
水产地93.7591.1593.7595.8394.2794.7996.3594.2795.8395.83
大棚57.1450.0057.1457.1471.4357.1450.0064.2957.1457.14
用户精度水稻74.1380.4374.1380.2977.2782.4478.0382.3183.3385.37
林地92.5691.1892.5691.9090.8291.7190.3488.8689.9590.87
廊道46.3953.0946.3952.8146.8457.5048.8653.4950.0054.12
菜地34.9448.3934.9443.9444.1654.6936.1148.2153.5764.00
水产地91.3788.3891.3793.4090.0591.4688.5294.7690.2092.00
大棚53.3363.6453.3357.1458.8257.1450.0069.2357.1461.54
表7  实验数据2中类别用户精度和制图精度 (%)
1 Song Qian, Zhou Qingbo, Wu Wenbin, et al. Recent Progresses in Research of Integrating Multi-source Remote Sensing Data for Crop Mapping[J]. Scientia Agricultura Sinica, 2015,48(6):1122-1135.
1 宋茜,周清波,吴文斌,等 农作物遥感识别中的多源数据融合研究进展[J]. 中国农业科学, 2015,48(6):1122-1135.
2 Du Baojia, Zhang Jing,Wang Zongming, et al. Crop Mapping based on Sentinel-2A NDVI Time Series Using Object-oriented Classification and Decision Tree Model[J]. Journal of Geo-information Science, 2019,21(5):740-751.
2 杜保佳,张晶,王宗明,等.应用Sentinel-2A NDVI时间序列和面向对象决策树方法的农作物分类[J].地球信息科学学报,2019,21(5): 740-751.
3 Han Yanxin, Meng Jihua. A Review of Per-field Crop Classification Using Remote Sensing [J].Remote Sensing for Land and Resources, 2019,31(2) :1-9.
3 韩衍欣,蒙继华.面向地块的农作物遥感分类研究进展[J]. 国土资源遥感,2019,31(2): 1-9.
4 Smith G M , Fuller R M. An Integrated Approach to Land Cover Classification: An Example in the Island of Jersey[J]. International Journal of Remote Sensing, 2001, 22(16):3123-3142. doi:10.1080/01431160152558288.
doi: 10.1080/01431160152558288
5 Lu D, Weng Q. A Survey of Image Classification Methods and Techniques for Improving Classification Performance[J]. International Journal of Remote Sensing, 2007, 28(5):823-870. doi:10.1080/01431160600746456.
doi: 10.1080/01431160600746456
6 Wang Dejun, Jiang Qigang, Li Yuanhua, et al. Land Use Classification of Farming Areas based on Time Series Sentinel-2A/B Data and Random Forest Algorithm[J]. Remote Sensing for Land and Resources, 2020, 32(4): 236-243.
6 王德军,姜琦刚,李远华,等.基于Sentinel-2A/B时序数据与随机森林算法的农耕区土地利用分类[J]. 国土资源遥感, 2020,32(4):239-246.
7 Deng Yuanyuan, Wu Zhaocong, Yi Lina, et al. Research on Object-oriented Classification of Agricultural Land based on High Resolution Images[J]. Remote Sensing for Land and Resources, 2010,22(4):120-124.
7 邓媛媛,巫兆聪,易俐娜,等.面向对象的高分辨率影像农用地分类[J]. 国土资源遥感, 2010,22(4):120-124.
8 José M, Barragán P, Ngugi M K, et al. Object-based Crop Identification Using Multiple Vegetation Indices, Textural Features and Crop Phenology[J]. Remote Sensing of Environment,2011,115(6):1301-1316. doi:10.1016/j.rse.2011. 01.009.
doi: 10.1016/j.rse.2011. 01.009
9 Nguyen-Thanh S, Chi-Farn C, Chen C, et al. Classification of Multitemporal Sentinel-2 Data for Field-level Monitoring of Rice Cropping Practices in Taiwan-Science Direct[J]. Advances in Space Research,2020,65(8):1910-1921. doi: 10.1016/j.asr.2020.01.028.
doi: 10.1016/j.asr.2020.01.028
10 Gu Xiaohe, Pan Yaozhong, He Xin, et al. Measurement of Sown Area of Winter Wheat based on Per-field Classification and Remote Sensing Imagery[J]. Journal of Remote Sensing , 2010, 14(4):789-805.
10 顾晓鹤,潘耀忠,何馨,等.以地块分类为核心的冬小麦种植面积遥感估算[J].遥感学报, 2010, 14(4):789-805.
11 Zhang B, Jia X , Chen Z , et al. A Patch-based Image Classification by Integrating Hyperspectral Data with GIS[J]. International Journal of Remote Sensing,2006,27(15):3337-3346. doi:10.1080/01431160500409577.
doi: 10.1080/01431160500409577
12 Kussul N, Lemoine G, Gallego F J, et al. Parcel-based Crop Classification in Ukraine Using Landsat 8 Data and Sentinel-1A Data[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2016, 9(6):2500-2508. doi: 10.1109/JSTARS.2016.2560141.
doi: 10.1109/JSTARS.2016.2560141
13 Lu Q, Huang X, Zhang L. A Novel Clustering-based Feature Representation for the Classification of Hyperspectral Imagery[J]. Remote Sensing,2014,6(6):5732-5753. doi:10.3390/rs6065732.
doi: 10.3390/rs6065732
14 Arthur D, Vassilvitskii S. K-Means++: The Advantages of Careful Seeding[C]// Proceedings of the Eighteenth Annual ACM-SIAM Symposium on Discrete Algorithms, SODA2007, New Orleans, Louisiana, USA, January7-9, 2007.
14 ACM, 2007.
15 He H, Bai Y, Garcia E A, et al. ADASYN: Adaptive Synthetic Sampling Approach for Imbalanced Learning[C]∥ Neural Networks, 2008. IJCNN 2008. (IEEE World Congress on Computational Intelligence). IEEE International Joint Conference on. IEEE, 2008.
16 Fan Dongdong, Li Qiangzi, Wang Hongyan, et al. Improvement in Recognition Accuracy of Minoritycrops by Resampling of Imbalanced Training Datasets of Remote Sensing[J]. Journal of Remote Sensing, 2019,23(4): 730-742.
16 樊东东,李强子,王红岩,等.通过训练样本采样处理改善小宗作物遥感识别精度[J].遥感学报, 2019,23(4): 730-742.
17 Tedros B,Charles L,Qiusheng W,et al.Decision-tree, Rule-based, and Random Forest Classification of High-resolution Multispectral Imagery for Wetland Mapping and Inventory[J]. Remote Sensing,2018,10(4):580. doi:10.3390/rs10040580.
doi: 10.3390/rs10040580
[1] 廖鸿燕,周小成,黄洪宇. 基于无人机遥感技术的台风灾害倒伏绿化树木检测[J]. 遥感技术与应用, 2021, 36(3): 533-543.
[2] 高子为,孙伟伟,程朋根,杨刚,孟祥超. 融合高分辨率遥感影像和POI数据的多特征潜在语义信息用于识别城市功能区[J]. 遥感技术与应用, 2021, 36(3): 618-626.
[3] 刘翔,樊海龙,郭建明,许世阳. 基于ASTER数据对中非裂谷系东部地表构造解释及油气信息提取[J]. 遥感技术与应用, 2021, 36(3): 649-662.
[4] 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.
[5] 王晨丞,王永前,王利花. 基于SAR纹理信息的农作物识别研究——以农安县为例[J]. 遥感技术与应用, 2021, 36(2): 372-380.
[6] 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.
[7] 马腾耀,肖鹏峰,张学良,马威,郭金金. 基于特征优选的GF-3全极化数据积雪识别[J]. 遥感技术与应用, 2020, 35(6): 1292-1302.
[8] 昝骁毓,谭晓悦,李强,陈晋. 融合灯光强度和斑块空间分布特征的贫困区域识别模型构建——以山西省为例[J]. 遥感技术与应用, 2020, 35(6): 1368-1376.
[9] 夏炎,黄亮,王枭轩,陈朋弟. 基于无人机影像的烟草精细提取[J]. 遥感技术与应用, 2020, 35(5): 1158-1166.
[10] 李萌,年雁云,边瑞,白艳萍,马金辉. 基于多源遥感影像的青海云杉和祁连圆柏分类[J]. 遥感技术与应用, 2020, 35(4): 855-863.
[11] 任慧敏,宋冬梅,王斌,甄宗晋,刘斌,张婷. 基于新极化特征参数的SAR海洋溢油检测[J]. 遥感技术与应用, 2020, 35(4): 934-942.
[12] 七珂珂,申茜,罗小军,李家国,姚月,杨崇. 基于GF-2影像的沈阳市黑臭水体遥感分级识别[J]. 遥感技术与应用, 2020, 35(2): 424-434.
[13] 边玲玲,王卷乐,郭兵,程凯,魏海硕. 基于特征空间的黄河三角洲垦利县土壤盐分遥感提取[J]. 遥感技术与应用, 2020, 35(1): 211-218.
[14] 张天媛,黄季夏,曹云锋,王利,孙宇晗,杨林生. 北极重要海峡海冰密集度时空变化呈现异质性[J]. 遥感技术与应用, 2019, 34(6): 1162-1172.
[15] 李强,冯德俊,瑚敏君,伍燚垚,杨历辉. 集成特征分量的高分二号影像阴影检测[J]. 遥感技术与应用, 2019, 34(6): 1252-1260.