联合相干性特征的短时序双极化SAR冬小麦提取  

Winter Wheat Extraction Based on Short Time-series SAR Data with Dual Polarizations and Coherence Characteristics

作　　者：郑希望 王凯[2] 朱野 嵇灵 王志勇[2,3] ZHENG Xiwang;WANG Kai;ZHU Ye;JI Ling;WANG Zhiyong(Levy Center of Natural Resources of Zhejiang Province,Hangzhou 310007,China;College of Geodesy and Geomatics,Shandong University of Science and Technology,Qingdao,Shandong 266590,China;National Demonstration Center for Experimental Surveying and Mapping Education(Shandong University of Science and Technology),Qingdao,Shandong 266590,China)

机构地区：[1]浙江省自然资源征收中心,杭州310007 [2]山东科技大学测绘与空间信息学院,山东青岛266590 [3]测绘工程国家级实验教学示范中心(山东科技大学),山东青岛266590

出　　处：《遥感信息》2024年第3期48-54,共7页Remote Sensing Information

基　　金：浙江省自然资源科技项目(2023-47)。

摘　　要：针对在多云多雨多雾地区光学遥感难以有效监测冬小麦的问题,构建了一种联合相干性特征的短时序双极化SAR冬小麦种植面积自动提取方法。该方法在HV/VV双极化SAR平均后向散射系数基础上,引入InSAR相干性特征,然后基于随机森林分类器获取冬小麦空间分布专题图,较好地解决乡间小道等被误分为冬小麦的问题。以山东省济宁市梁山县为实验区,利用20景Sentinel-1A双极化SAR数据,获取了梁山县2022年的冬小麦种植面积及空间分布。结果表明:文章构建的基于短时序双极化SAR数据的冬小麦提取方法,总体精度为92.799%,Kappa系数为0.912,与未引入相干性特征相比,总体精度提升6.279%,Kappa系数提高0.096;可以利用抽穗期或开花乳熟期的短时序SAR数据代替完整生长期的时序SAR数据,总体精度相差分别为3.008%和3.341%。To solve the problem that it is difficult to effectively monitor winter wheat in cloudy,rainy and foggy areas using the optical remote sensing data,an automatic method of extracting the planting area of winter wheat based on short time-series SAR data with dual polarizations is proposed.Based on the mean backscattering coefficient of HV/VV dual polarizations,InSAR coherence feature is introduced to establish a suitable classification feature dataset for extracting the winter wheat,which can better solve the problem of misclassification of rural trails.Then,the spatial distribution thematic map of the winter wheat in the study area is obtained based on random forest classifier.Taking Liangshan county,Shandong province as an experimental area,the planting area and spatial distribution of winter wheat are obtained by using 20 Sentinel-1A SAR data with dual polarizations from October 2021 to June 2022.The results show that:the method based on the short time-series SAR data could meet the requirements of winter wheat monitoring,with the overall accuracy reaching 92.799%and the Kappa coefficient reaching 0.912;compared with no coherence feature,the overall accuracy is improved by 6.279%,and the Kappa coefficient is increased by 0.096;the short time-series SAR data of tassel stage can be used to replace the time-series SAR data of the full growth period,and the data calculation amount is greatly reduced,while the accuracy difference is only 3.008%;the second option is flowering and milking stage,and compared with the full growth period,the overall accuracy difference is only 3.341%.

关 键 词：短时序SAR 冬小麦 双极化 相干性 随机森林 

分 类 号：TP722.6[自动化与计算机技术—检测技术与自动化装置]