基于NDVI-NSSI空间与HSV变换的成熟期农作物遥感识别  

Crop Identification in Mature Stage with Remote Sensing Based on NDVI-NSSI Space and HSV Transformation

作　　者：宋承运[1,2] 曲雪杉 胡光成 苏涛 SONG Chengyun;QU Xueshan;HU Guangcheng;SU Tao(School of Geomatics,Anhui University of Science and Technology,Huainan 232001,China;Key Laboratory of Aviation-Aerospace-Ground Cooperative Monitoring and Early Warning of Coal Mining-induced Disasters of Anhui Higher Education Institutes,Huainan 232001,China;State Key Laboratory of Remote Sensing Science,Aerospace Information Research Institute,Chinese Academy of Sciences,Beijing 100101,China)

机构地区：[1]安徽理工大学空间信息与测绘工程学院,淮南232001 [2]矿山采动灾害空天地协同监测与预警安徽普通高校重点实验室,淮南232001 [3]中国科学院空天信息创新研究院遥感科学国家重点实验室,北京100101

出　　处：《农业机械学报》2023年第8期193-200,共8页Transactions of the Chinese Society for Agricultural Machinery

基　　金：矿山采动灾害空天地协同监测与预警安徽普通高校重点实验室(安徽理工大学)开放基金项目(KLAHEI202205)。

摘　　要：成熟期农作物的识别在农作物种植面积估算、农业生产及产量统计方面具有重要作用。为提供一种简便的成熟期农作物遥感识别方法,利用Sentinel-2A数据,以安徽省滁州市凤阳县为研究区,通过归一化植被指数(Normalized difference vegetation index,NDVI)与归一化光谱分离指数(Normalized spectral separation index,NSSI)构成的空间,提取光合植被、非光合植被、裸土的纯端元,由像元三分模型,得到非光合植被覆盖度及成熟期农作物的空间分布。为进一步提取研究区内具有相同成熟期的冬小麦与油菜,利用油菜开花期Sentinel-2A数据,由Hue saturation value(HSV)图像变换方法,分别提取出成熟期冬小麦与油菜。与地面观测数据和辅助数据相比,提取的成熟区冬小麦、油菜的总体精度为95.34%,Kappa系数为0.904,高于支持向量机方法(总体精度91.66%,Kappa系数为0.813)与决策树方法(总体精度92.39%,Kappa系数为0.838)的提取精度。结果表明,NDVI-NSSI空间与HSV变换相结合的方法,可以有效将非光合植被与土壤背景分离,识别成熟期冬小麦与油菜,具有对数据需求较少,易操作等优势,也为提取农作物成熟期内的裸地以及与裸地具有相似波谱的地物提供了思路与方法。Identification of mature crops plays an important role in crop area estimation,agricultural production,and yield statistics.To provide a simple method for identification of crop at mature stage with remote sensing,the method based on normalized difference vegetation index(NDVI)-normalized spectral separation index(NSSI) space and Hue saturation value(HSV) transformation was proposed by using Sentinel-2A data.Fengyang County located in Chuzhou City,Anhui Province,was selected as study area.The pure pixel of photosynthetic vegetation,non-phototrophic vegetation and bare soil were estimated by using the NDVI-NSSI space firstly.Coverage of non-phototrophic vegetation and the distribution of mature crops were estimated by using spectral mixture analysis method.As wheat and rape had the same maturity period in the study area,HSV transformation was used to estimate the distribution of rape by using the Sentinel-2A data at the flowering stage of rape,and the mature wheat and rape was identified.The results were compared with ground observation data and auxiliary data,the accuracy of wheat and rape was 95.34%,and the Kappa coefficient was 0.904,which was higher than that of support vector machine method(accuracy was 91.66%,Kappa coefficient was 0.813) and the decision tree classification method(accuracy was 92.39%,Kappa coefficient was 0.838).The results indicated that non photosynthetic vegetation could be separated from soil background areas by using the NDVI-NSSI space and HSV transformation,and mature wheat and rape could be identified.The method can be used with less data demand and easy operated,which can provide idea and method for extracting bare soil during crop maturity and features with similar spectra to bare soil.

关 键 词：成熟期农作物 遥感识别 非光合植被 NDVI-NSSI空间 混合像元分解 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]