基于无人机和地面数字影像的水稻氮素营养诊断研究

选用扫描仪和无人机平台获取水稻叶片和冠层的数字图像,运用数字图像处理技术研究不同氮素营养水平水稻叶片和冠层的综合特征信息,从而应用于水稻的氮素营养诊断。结果表明:1)通过叶片叶绿素a含量和扫描叶片颜色参量之间的相关性分析,得到可用于诊断水稻氮素营养水平的叶片颜色特征参量B、b、b/(r+g)、b/r、b/g。通过叶片的颜色、形状综合特征信息与YIQ电视信号彩色坐标系统的参量建立氮素营养的识别模型,4个不同氮素水平的正确识别率分别为:N0(0 kg N.hm-2)74.9%,N1(60 kgN.hm-2)52%,N2(90 kg N.hm-2)84.7%,N3(120 kg N.hm-2)75%;2)无人机获取的田间冠层图像识别水稻氮素营养水平的综合特征参量是G、B、b、g、b/(r+g)、b/r、b/g、H、S、DGCI,选择相同的CB参量建立冠层氮素营养的识别模型,4个不同氮素水平的正确识别率为:N0(0 kg N.hm-2)91.6%,N1(60 kgN.hm-2)70.83%,N2(90 kg N.hm-2)86.7%,N3(120 kg N.hm-2)95%。初步研究表明基于综合特征的氮素诊断模型区分效果比较好,利用叶片扫描图像和无人机识别与诊断田间水稻氮素是可行的。

Diagnoses of rice nitrogen status based on spectral characteristics of leaf and canopy

The scanner and unmanned aerial vehicle(UAV) were adopted to take the leaf and canopy images,respectively.The images were corrected and applied to assess the rice nitrogen status on leaf level to canopy level by image analysis.The main results were as follows: 1) According to the analysis of relationship between the content of chlorophyll a and the color parameters,the effective color parameters were discovered as B,b,b/(r+g),b/r,b/g.The regression was carried out on leaf level based on the distinctive char...