基于高分辨率遥感影像提取的种植结构信息，能够比传统的统计数据更加直观地表达农作物的空间分布特征，这些数据信息是水资源管理部门进行水资源管理的重要数据参考。本文为解决GF-1 WFV传感器影像中混合像元对小麦信息提取结果的影响，引入高分辨率GF-1 PMS传感器影像，在两种影像相同位置建立样本研究区，利用PMS影像的分辨率优势为WFV影像中小麦混合像元训练样本提供真实小麦面积权重，得到WFV影像小麦混合像元NDVI与小麦面积权重的比例关系，再运用区间归一化的方法解决同一NDVI值对应不同小麦面积权重的问题，进而得到混合像元中小麦的真实面积信息，最终提取了冀州市的冬小麦信息。经验证，该方法能够在实地样本不足的条件下，较准确地提取冬小麦面积信息。

The planting structure information extracted by high-resolution remote sensing imaging can be more intuitive than traditional statistical data to present the spatial distribution and area information of crops. These data can provide important reference for water resources management. In order to eliminate the influence of mixed pixels in GF-1 WFV sensor images on wheat information extraction, we introduced the high-resolution GF-1 PMS sensor images, and established samples in the two images. We used the superior resolution of PMS images to provide real wheat area weight to the training samples of wheat mixed pixels in the WFV image, and obtained the relationship between wheat NDVI and wheat area weight. Then we used the interval normalization method to solve the problem in which one NDVI value corresponded to different wheat area weights, and thus obtained the true area information of wheat in mixed pixels, and extracted the winter wheat information of Jizhou City. It was verified that the method can accurately extract information of winter wheat area under the condition of insufficient samples.

国家科技重大专项（08Y30B07-9001-13/15）；国家科技支撑计划（2013BAB05B01）