基于多光谱图像的水稻叶片叶绿素和籽粒氮素含量检测研究

Measurement of Rice Leaf Chlorophyll and Seed Nitrogen Contents by Using Multi-Spectral Imagine

先利用常规技术分析了水稻的叶片叶绿素和籽粒氮素含量,然后用包含绿(G)、红(R)和近红外(NIR)三波段通道的电荷耦合器件(CCD)成像技术对水稻叶片和籽粒进行了无损检测。试验结果显示,水稻叶片叶绿素a、叶绿素b分别与G、NIR通道图像灰度呈极显著线性相关,叶绿素(a+b)含量则与上述两通道图像灰度呈显著线性相关;而且,水稻籽粒氮素含量与G、NIR通道、归一化植被指数(NDVI)灰度呈显著线性相关。由此建立了水稻叶片叶绿素和籽粒氮素含量的多光谱图像预测模型,并分别用21个样本对模型进行检验,其中线性显著相关的7个模型的相对误差RE(%)介于9.36%～15.7%,实现了对水稻叶片叶绿素和籽粒氮素含量的快速、准确、非破坏性检测。

To determine rice leaf chlorophyll and seed nitrogen contents, a multi-spectral sensor which assesses the biochemical content of rice by means of gray values sensed using three channels （green, red, near-infrared） of the multi-spectral camera was used. The results showed that there were extremely significant correlations between the chlorophyll a content, chlorophyll b content in leaves and the gray values of green channel, near infrared channel respectively and significant correlation between the chlorophyll （a＋b） content in leaves and the gray values of green channel, near-infrared channel. Similarly, there was a significant correlation between the seed nitrogen content and the gray values of green channel, near-infrared channel and normalized difference vegetation index. Moreover, regression equations between gray values of multi-spectral imagine and leaf chlorophyll content or seed nitrogen content were verified with 21 samples and the relative error of 7 models ranged from 9. 36% to 15. 7%. Thus, the rapid, accurate and non-destructive estimations of leaf chlorophyll and seed nitrogen contents were realized.