【摘要】 针对难以对农作物收割过程进行有效地遥感监测这一难题,采用时空数据融合模型重构出空间分辨率为30m,时间分辨率为1d的高时-空分辨率遥感数据对农作物收割进度进行监测。针对冬小麦收割前后NDVI变化呈现的线性特征,计算时序NDVI的曲率,通过曲率确定阈值进行收割信息提取。结果表明:采用面向对象的SVM分类方法提取研究区冬小麦种植信息,Kappa系数为0.901,面积误差为2.61%;时空数据融合模型的结果与真实影像间的相关系数为0.77(红波段)和0.78(近红外波段),能够较好地重构出冬小麦收割时期的影像;通过提取NDVI变化曲线曲率接近于0时的冬小麦NDVI值来确定阈值,实现了冬小麦收割过程信息的遥感提取。更多还原

【Abstract】 To solve the technology problem of monitor the harvesting process by remote sensing,we chose the spatial and temporal data fusion approach(STDFA)model to rebuild the remote sensing data which has 30 meter spatial and one day temporal resolution and then the reconstructed data set were used to extract the crop harvesting information.According to the linear variation of NDVI value before and after harvest of winter wheat,we calculated the curvature of time-series NDVI dataset and the threshold was determined by the curvature of NDVI value.The results show that:the Kappa index is 0.901 and the area error is 2.61%though the object based support vector machine(OB-SVM)classification method;the correlation coefficient of the result of STDFA and true reflectance in red band and near infrared band is 0.77 and 0.78,which means that the spatiotemporal fusion method can rebuild the data during the harvesting time of winter wheat;the threshold is determined by extracting the NDVI of winter wheat pixel which has a curvature value close to 0,and then the harvesting information about winter wheat is extracted by using the remote sensing technology.更多还原