摘要
充分利用高空间分辨率和多时相遥感影像优势,提出一种时空影像融合的灌区作物提取方法。以河套灌区为例,通过分析不同作物物候差异及其在Sentinel-2卫星影像的归一化植被指数(NDVI)时序数据的特征表现,优化选取作物物候差异大的时相组合,利用Gram-Schmidt影像融合方法对优选影像与2 m高空间分辨率全色影像进行图像融合,生成时空融合影像。基于时空融合影像进行面向对象图斑分割,结合作物物候差异分析构建分类规则并进行河套灌区作物提取,评价分类精度。结果表明,时空遥感影像融合技术可以增强灌区作物识别能力,分类总体精度为94.8%,Kappa系数为0.921,可为灌区用水精细化管理提供基础数据。
To address insufficient of high spatial resolution and multi⁃temporal remote sensing image fusion in crop classification,a spatial⁃temporal image fusion method was presented for crop identification of irrigation area.In this paper,taking Hetao irrigation area as an exam⁃ple,the phenology differences of different crops were analyzed,combining its performance in normalized difference vegetation index(NDVI)time⁃series data of Sentinel 2.Then three temporal images with large differences in crop phenology were selected to fuse with 2 m spatial reso⁃lution panchromatic images by using Gram⁃Schmidt image fusion method.The fused high spatial resolution NDVI images were stacked to gen⁃erate spatial⁃temporal fusion images.Based on image segmentation of spatial⁃temporal fusion images,cropland types were extracted by using decision tree classification.The proposed method of crop planting structure extraction obtained high accuracy with an overall accuracy of 94.8%and a Kappa coefficient of 0.921.
作者
陈亮
杨阳
宋伟
卢欣
申源
CHEN Liang;YANG Yang;SONG Wei;LU Xin;SHEN Yuan(Information Center of Yellow River Conservancy Commission,Zhengzhou 450004,China)
出处
《人民黄河》
CAS
北大核心
2022年第12期154-157,162,共5页
Yellow River
基金
黄委信息中心创新团队项目(HWXXZXCX-202201)。
关键词
时空融合影像
面向对象
作物
NDVI
河套灌区
spatial⁃temporal fusion image
object⁃oriented
crop
NDVI
Hetao irrigation area
