基于3-D数据的叶片萎蔫形态辨识方法

Identification Methods for Plant Wilting at Leaf-scale Based on 3-D Imaging

以西葫芦为试验对象,采用非接触式激光扫描获取植物叶片3-D形态信息,运用微分几何算法、二维傅里叶谱分析法、垂直投影叶面积法以及标准差法定义了4种植物萎蔫指数(TB_3D、RLWI-2DFT、RTPLA和RSD)用来定量刻画植物萎蔫状态。试验结果表明萎蔫指数能够很好地定量刻画植物萎蔫形态。其中TB_3D与太阳全辐射量、大气温度线性相关,决定系数取值范围分别为[0.710,0.721]和[0.513,0.589]。TB_3D与茎秆直径线性相关,决定系数为0.845。最后通过比较4种萎蔫指数刻画萎蔫状态的相关性,两两间决定系数大于0.736,结果表明4种指数的萎蔫表征效果基本一致。

To characterize plant wilting status, a laser scanner based on triangulation principle was used to obtain 3-D images of zucchini. Four wilting indices （ TS_3D, RLWI.2DFT, RTPLA, RSD ） were developed by using different mathematic methods （differential geometry, 2-D Fourier transform, top projected area, standard deviation） to identify the plant wilting. It was observed that the indices were capable of sensing the stress response of zucchini from slight wilting to severe stress levels. Regression results showed that TB3D had a linear relationship with ambient temperature and global radiation, the decision coefficient were [0. 710, 0. 721 ] and [0. 513, 0. 589], respectively. Besides, TB3D was also found to be linearly correlated with stem diameter, and R2 was 0. 845. Additionally, the correlations of wilting series characterizing by these four wilting indices were analyzed respectively. The R2 of each two indices was larger than 0. 736. The results showed that the four wilting indices had the similar effects.