【摘要】 利用精米粉脂肪含量化学分析值及其近红外反射光谱以建立分析精米中脂肪含量数学模型。选取 3段谱区(6 10 0～4 2 5 0cm- 1 ;6 10 0～ 5 4 5 0cm- 1 并 4 6 0 0～ 4 2 5 0cm- 1 ;75 0 0～5 4 5 0cm- 1 并 4 6 0 0～4 2 5 0cm- 1 )分别采用 4种不同的预处理方法(直线扣除、乘性散射校正、矢量归一、一阶导数 ) ,对其结果进行比较 ,发现选择不同谱区和不同光谱预处理方法对所建模型有很大的影响 ,同时证明了利用近红外光谱技术测定稻米脂肪含量的可行性。还阐述了如何对模型进行评价 ,并讨论了它在遗传育种中的应用。更多还原

【Abstract】 Three spectrum regions (6100-4250 cm -1, 61 00 -5450 cm -1 and 4600-4250 cm -1, 7500-5450 cm -1and 4600-4250 cm -1) and four pretreatment methods (straight line subtraction, multiplicati ve scattering correction, vector normalization; first derivation ) were compare d to form a model for testing fat content in milled rice by using near infrared reflectance spectroscopy (NIRS). It was found that spectrum regions and pretreat ment methods had large effects on the model construction. In order to estimate the forecast result of the model, the feasibility of testing fat content in mil led rice by using NIRS and its application in breeding was discussed.更多还原