基于ROC图的油菜生长期光谱敏感波段的研究
DOI:
CSTR:
作者:
作者单位:

作者简介:

通讯作者:

中图分类号:

基金项目:

国家自然科学基金项目(31501227,11571103);湖南省科技重大专项(2014FJ1006);湖南省科学技术厅重点研发计划项目(2015JC3098);湖南省教育厅重点项目(15A083);湖南农业大学大学生科技创新基金项目(2016ZK32)


Author:
Affiliation:

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    测定了直播和移栽油菜的苗期、抽薹期、花期、盛花期和角果期的冠层光谱,构建了比值光谱植被指数(RSI)和归一化光谱植被指数(NDSI)。为了获得区分直播和移栽的最佳RSI和NDSI,利用降采样法和精细采样法相结合的受试者工作特征(ROC)图寻找油菜生长期光谱的敏感波长,直播和移栽油菜各时期RSI和NDSI的最敏感波长分别为:苗期(458 nm,511 nm)和(433 nm,517 nm);抽薹期(997 nm,501 nm)和(990 nm,510 nm);花期(1 235 nm,1 180 nm)和(1 235 nm,1 180 nm);盛花期(478 nm,396 nm)和(484 nm,416 nm);角果期(1 073 nm,1 037 nm)和(1 092 nm,1 024 nm)。用敏感波长下的2种植被指数为特征,以最近邻法为分类器的定性识别模型,结果花期的区分效果最好,最大约登指数分别为0.941 7和0.945 0。

    Abstract:

    In this paper, rapeseed canopy spectra was collected, in periods of seeding, bolting, flowering, lush flowering and pod, to generate ratio spectral index (RSI) and normalized difference spectral index (NDSI). In order to acquire the best RSI and NDSI, the most sensitive bands were obtained by employing receiver operating characteristic (ROC) map with the combination of down sampling and fine sampling. For the direct plant and transplant the best values are as followed: the seeding period: RSI(458 nm, 511 nm) and NDSI(433 nm, 517 nm), the bolting period: RSI(997 nm, 501 nm) and NDSI(990 nm, 510 nm), the flowering period: RSI(1 235 nm, 1 180 nm) and NDSI(235 nm, 1 180 nm), the lush flowering period: RSI(478 nm, 396 nm) and NDSI(484 nm, 416 nm), and the pod period: RSI(1 073 nm, 1037 nm) and NDSI(1 092 nm, 1 024 nm), respectively. By using the two best RSI and NDSI as features and the nearest neighbor method as classifier, the results of the qualitative identification model show that the best discrimination effect comes from the flowering period, the corresponding Youden index is 0.941 7 and 0.945 0, respectively.

    参考文献
    相似文献
    引证文献
引用本文

段少新,姜珊,王访,邹锐标,廖桂平.基于ROC图的油菜生长期光谱敏感波段的研究[J].湖南农业大学学报:自然科学版,2017,43(6):.

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2017-12-11
  • 出版日期:
文章二维码