中文摘要：

应用可见／近红外漫反射光谱对梨表面色泽进行无损检测研究．在350～1800nm光谱区间，结合梨的原始吸收光谱和标准化光谱，采用多元线性回归（MLR）、主成分回归（PCR）和偏最小二乘回归（PLSR）三种数学校正算法进行了定量对比分析．原始吸收光谱应用偏最小S-乘回归建立的定标模型对24个未知样品的预测结果是：L^＊、a^＊、b^＊预测均方差分别为1．4251，0．4569和0．9497；相对预测偏差分别为3．7404％，3．3571％和2．5877％．实验结果表明：可见／近红外光谱技术对梨表面色泽的无损检测具有可行性．

英文摘要：

Visible/near-infrared （VIS/NIR） diffuse reflectance spectra were applied to non-destructive measurement of surface color of pear. In the wavelength range of 350 - 1800nm, multi-linear regression （MLR）, principal component regression （PCR） and partial least squares regression （PLSR）, combining with the original absorbance spectra（log（1/R） ） and its standardized spectra （1/SD）, were used to quantitatively analyze surface color of pear. The prediction results for root mean standard error of prediction（RMSEP） and relative standard deviation RSD% of 24 unknown samples, are 1. 4251 and 3. 7404% for L^＊, 0.4569 and 3.3571% for a^ ＊, 0. 94.97 and 2.5877% for b^ ＊ , based on calibration model with PLSR by using log（I/R）. The experiment results sbow tbat non-destructive measurement of surface color of pear by using VIS/ NIR spectroscopy technique is feasible.