基于高光谱技术的基质含水率快速测定方法

Measuring Moisture Content of Rice Husk Using Hyperspectral Technology

为了寻找一种基质含水率的快速检测方法,应用高光谱技术获得不同含水率基质样品光谱信息,阐释基质含水率光谱规律,对基质含水率进行定量分析,为设施基质栽培水分快速测定提供参考。以稻壳基质为研究对象,对基质光谱信息进行基线校正和平滑处理后,利用逐步回归分析法提取稻壳基质光谱反射率一阶微分变换的敏感波段,建立基于敏感波段组合的基质含水率预测模型,并对模型进行了检验。结果表明,在敏感波段527、796和959 nm处,采用反射率一阶微分建立的稻壳基质含水率三波段指数预测模型的预测效果较好,模型预测相关系数(RP)为0.91,预测均方根误差(RMSEP)为5.55%,可以实现对稻壳基质含水率的快速准确检测。

We took rice husk as an example, examining the suring moisture content in organic substrates. The spectra potential of hyperspectral technology for quickly mea- of the substrate samples with different moisture con- tents was measured by the hyperspectral imaging technology, and we then calculated the most sensitive wave- lengths by transforming the first-order derivative of the reflectivity derived using the stepwise linear regression, from which we established the relationship between the wavelength and the moisture content. The results showed that the characteristic wavelength was 527 nm, 796 nm and 959 nm. It was found that the first-order derivative of the reflectivity in the three wavelengths can predict moisture content in the substrate with a correlation coefficient of 0.91 and a root mean squared error of 5.55%. The hyperspectral technology can thus be used as a reliable method for rapid measurement of moisture content in rice husk.