脐橙放置方位对近红外光谱检测结果的影响

研究了不同放置方位对近红外光谱检测脐橙可溶性固形物(soluble solids content, SSC)结果的影响。按照脐橙果梗、窝底与入射光线所成角度的不同，分成垂直(90°)、平行(0°)和任意(不包含0°和90°)三种放置方位，并采用USB4000微型光纤光谱仪在465～1 150 nm范围内以半透射方式分别采集三种放置方位的脐橙光谱。脐橙样品数为336个，其中228个样本作为校正集，其余108个样本作为预测集。应用偏最小二乘回归并结合不同的预处理方法建立不同方位下的脐橙SSC预测模型，结果表明果梗、窝底与光源所成角度为垂直时脐橙SSC预测模型性能较优，其最好的模型相关系数r_c=0.93，RMSEC=0.37%，r_p=0.88，RMSEP=0.49%。

Near-Infrared Spectrum Detection Result Influenced by Navel Oranges Placement Position

The present paper studies the near-infrared spectroscopy of soluble solids content in navel orange that was influenced by different placement positions of the navel orange. According to the different angles between incident light and the straight line composed by navel orange stems and pit, the authors chose three different placement positions, vertical (90°), parallel (0°) and random(not including 0°and 90°). The authors acquired the semi-transmission spectrum of the navel orange placed in different positions in the wavelength range of 465～1 150 nm by the miniature fiber spectrometer USB4000, there were 336 navel orange samples in the experiment, 228 samples weree used as calibration set, and the rest 108 samples were used as prediction set. The authors used partial least-square regression combined with different pre-processing methods to establish the prediction model of SSC in navel orange with different placement positions. The result shows that when the angle is vertical the prediction models of SSC in navel orange are good, and the best correlation coefficient of the model is r_c=0.93，RMSEC=0.37%，r_p=0.88，and RMSEP=0.49%.