Since remote sensing technology based on optical images is usually influenced by cloud and rain,it's difficult to acquire continuous crop growth curves in some areas.Radar,as an active remote sensing technique,can overcome the disadvantage successfully.Taking the farm located in the city of Weinan of Shaanxi Province as study area,two methods of maximum likelihood (ML) and support vector machine (SVM) were adopted to combine multi-sensor remote sensing data of Sentinel-1 and Sentinel-2,and thus improve crop classification accuracy.The results showed that classification results with fusion data were better than those of optical data.The classification result of fusion data composed of Sentinel-1 and Sentinel-2's red,green,blue and near-infrared bands with no cloud were improved evidently with SVM method.The overall accuracy and Kappa coefficient were raised by 2 percentage points and 5 percentage points,respectively.In the case of a few cloud cover in the study site,the overall accuracy and Kappa coefficient with ML method were increased by 2 percentage points and 4 percentage points,respectively.With SVM method,the overall accuracy and Kappa coefficient were raised by almost 6 percentage points and 8 percentage points,respectively.%基于光学影像的遥感技术受云雨、昼夜影响较大,导致获取连续的作物时序生长曲线较困难,而雷达影像作为主动式成像,能够很好地克服这一缺陷.本文以陕西省渭南市大荔县某农场为研究区域,分别采用最大似然法(Maximum likelihood,ML)和支持向量机(Support vector machine,SVM)2种方法,融合Sentinel-1雷达影像和Sentinel-2光学影像,提高农作物的分类精度.研究结果表明,融合数据的农作物分类精度相比光学数据分类精度有所提高.在无云层覆盖的情况下,利用SVM方法融合Sentinel-2的红、绿、蓝、近红外4个波段数据与Sentinel-1数据,总体分类精度提高了2个百分点,Kappa系数提高了5个百分点;在有少量云层覆盖情况下,利用ML处理融合数据的分类结果精度和Kappa系数分别提高2个百分点和4个百分点,SVM方法下的分类精度提高了6个百分点,Kappa系数提高了8个百分点.
展开▼
