HYPERSPECTRAL ESTIMATION MODELS FOR CRUDE FIBRE CONCENTRATION OF CORN
玉米粗纤维含量高光谱估算模型研究

The crude fibre concentration of leaf, stem, ear and sheath of six corn cultivars (Zea mays L.) and their corresponding laboratory spectral reflectance were measured. By correlation analysis, the most sensitive wavelengths for raw spectral reflectance and the first derivative spectral reflectance for crude fibre estimation were located at 1924nm and 2370nm, respectively. The linear and nonlinear fitting analyses indicate that the linear model based on the first derivative spectral reflectance at 2370nm is the most suitable one for the estimation of crude fibre concentration. Compared to the other three models, the linear model has the smallest RMSE(RMSE=3.538), and the theoretical curde fibre concentration derived from the linear model is significantly correlated with the measured crude fiber concentration, and the correlation coefficient r=0.693.