论文部分内容阅读
通过测试不同试验条件下稻米品质及其高光谱数据,研究稻米中淀粉、蛋白质和胶稠度与高光谱数据特征之间的相关关系。结果表明:在不同施肥条件下,通过构建归一化指数(NDI,x)与稻米品质指标(y)的多种关系模型,比较模型预测的显著性,得出的多项式模型对稻米品质指标具有较高预测性,通用模型y=ax2+bx+c中的构成系数a、b、c因肥料水平差异取值不同;在不同品种条件下,通过比较波段比值指数(RI,x)与稻米品质指标(y)的相关系数,得出的最佳波段组合分别为x直链淀粉=R783nm/R634nm、x蛋白质=R829nm/R646nm和x胶稠度=R900nm/R670nm,推算出预测稻米品质指标的线性回归方程通用模型y=ax+b中的构成系数a、b因品种差异取值不同。这表明,运用高光谱方法估算稻米中的直链淀粉、蛋白质含量和胶稠度切实可行,可为稻米品质遥感监测提供依据。
By testing the rice quality and its hyperspectral data under different experimental conditions, the correlation between the starch, protein and gel consistency and the characteristics of hyperspectral data in rice was studied. The results showed that under different fertilization conditions, the model predictions were compared by constructing multiple relational models between normalized index (NDI, x) and rice quality index (y). The polynomial model obtained had the following characteristics Higher predictiveness, the constituent coefficients a, b and c in the general model y = ax2 + bx + c are different due to the different values of the fertilizer levels. By comparing the ratio index (RI, x) (Y), the best combination of bands was derived as follows: x amylose = R783nm / R634nm, xprotein = R829nm / R646nm and x gel consistency = R900nm / R670nm, Common factors in the equation y = ax + b in the constituent coefficients a, b due to differences in the value of different species. This shows that it is practicable to estimate the amylose, protein content and gel consistency in rice by using hyperspectral method, which can provide the basis for remote sensing monitoring of rice quality.