基于光谱和时相特征的夏玉米遥感识别

发布时间：2018-10-22 09:12

【摘要】：精确提取作物种植面积一直是农业遥感关注的主要问题之一。综合运用低分辨率的时相变化特征和中分辨率的光谱特征,提出一种夏玉米识别方法。首先基于MODIS NDVI时间序列曲线,分析夏玉米在时相变化上的识别特征,构建识别模型。夏玉米纯像元利用识别模型识别,而耕地和非耕地类型的植被产生的混合像元,则基于像元分解办法获取耕地组分的NDVI时序特征,再利用识别模型判定,然后结合土地利用数据根据空间关系得到中分辨率结果;玉米与其他作物的混合像元则利用中分辨率尺度光谱差异加以区分。研究结果表明,在伊洛河流域主要农业区,识别精度达到90.33%,为作物类型识别提供了新的思路。
[Abstract]:Accurate extraction of crop planting area has been one of the main problems of agricultural remote sensing. A recognition method for summer maize was proposed by using the characteristics of low resolution temporal variation and medium resolution spectrum. Firstly, based on the MODIS NDVI time series curve, the recognition characteristics of summer maize in temporal change were analyzed and the recognition model was constructed. The pure pixels of summer maize are identified by the recognition model, while the mixed pixels produced by the vegetation of cultivated land and non-cultivated land are used to obtain the NDVI temporal characteristics of cultivated land components based on the method of pixel decomposition, and then the recognition model is used to judge. Combined with land use data, the meso-resolution results are obtained according to spatial relationships, and the mixed pixels of maize and other crops are distinguished by mesoscale spectral differences. The results show that the recognition accuracy is 90.33 in the main agricultural areas of the Yiluo River basin, which provides a new idea for crop type identification.
【作者单位】：黄河水利职业技术学院;黄河中下游数字地理技术教育部重点实验室;
【基金】：河南省科技厅科技攻关项目(152102110047、142102110098) 河南省教育厅科学技术研究重点项目(13A420617) 中国博士后科学基金资助项目(20100470994)
【分类号】：S513;S127

【参考文献】