重构玉米叶片光谱的铜铅污染区分与其含量反演方法研究

利用经验模态分解(empirical mode decomposition,EMD)、自相关函数一阶导数比值差(ratio difference of autocorrelation function first-order derivative,_(RDA))和多重分形理论,结合玉米盆栽实验,研究了铜(Cu~(2+))、铅(Pb~(2+))离子不同胁迫梯度下玉米叶片光谱去噪、叶片重金属污染的Cu和Pb元素区分以及叶片中Cu、Pb元素含量预测方法。通过光谱数据的EMD去噪与重构处理,得到不同浓度Cu、Pb胁迫下玉米叶片重构光谱;利用光谱自相关函数一阶导数(autocorrelation function first derivative,AFFD)及其比值差(_(RDA)),建立了Cu~(2+)、Pb~(2+)不同胁迫梯度下玉米叶片重构光谱的_(RDA)变化量(Cu_(RDA)、Pb_(RDA))计算公式;依据_(RDA)变化量曲线中紫光、绿峰、红光、红边、近谷、近峰多个波谱特征区间的Cu_(RDA)和Pb_(RDA)计算值,可明显地区分出叶片的Cu、Pb污染类别;另外,根据实测的玉米叶片中叶绿素、Cu~(2+)、Pb~(2+)含量与叶片重构光谱的多重分形谱参量之间相关性,构建了叶片中Cu~(2+)、Pb~(2+)含量反演的线性回归预测模型,经验证模型精度较高。

A Study of Copper-Lead Pollution Distinctions and a Content Retrieval Method for the Reconstruction of Maize Leaf Spectra

In this study, a de-noising method of corn leaf spectra, which had been stressed by heavy metals of different concentration gradients, was discussed. Also, the distinctions between the Cu and Pb element levels in the contaminated leaves, and a prediction method for the Cu2+ and Pb2+ content in the leaves based on Empirical Mode Decomposition (EMD), Ratio Difference of Autocorrelation Function First-order Derivative (RDA), and a multi-fractal theory combined with potted corn plant experiments, were utilized. An EMD de-noising, as well as reconstruction processes of the spectral data, were adopted to achieve the reconstructed spectra of the corn leaves stressed by different concentration gradients of Cu2+ and Pb2+. Also, an autocorrelation function first-order derivative (AFFD), and the RDA of the spectra, were adopted to establish the RDA variation (CuRDA and PbRDA) calculation formula of the corn leaves which had reconstructed spectra stressed by different concentration gradients Cu2+ and Pb2+. Then, in accordance with the CuRDA and PbRDA calculated values of the multiple spectral characteristic intervals (purple, green-peak, red, red-edge1, red-edge2, red-edge3, red-edge4, near-valley, near-peak B1, and near-peak B2) in the RDA variation curve, this study was able to clearly distinguish the Cu and Pb pollution categories of the leaves. In addition, this study established a linear regression forecasting model, along with a Newton interpolation polynomial forecasting model, for the content levels of the Cu2+ and Pb2+ inversion based on the correlation coefficient between the measured chlorophyll, Cu2+, and Pb2+ content, and the multi-fractal spectrum parameters of the reconstructed spectra of the leaves.