Abstract Paprika powder is a widely consumed spice, making it an attractive target for adulteration, which is not easily detected. In this study, a portable near-infrared (NIR) spectrometer was used for fast detection of paprika adulteration. Nine paprika samples from five suppliers were adulterated with potato starch, acacia gum and annatto at different concentrations (0–36% by weight of potato starch and acacia gum, and 0–18% by weight of annatto). The NIR spectrum of each mixture (n = 315) was used as predictors to determine adulteration by partial least squares-discriminant analysis (PLS-DA) and partial least squares regression (PLSR). First, PLS-DA was applied to discriminate between adulterated and non-adulterated samples, as well as the type of adulterant. This method proved to be efficient, with specificity greater than 90 % and error rate lower than 2 %, for all models constructed. PLSR was used to predict the concentration of adulterants in paprika samples. In addition, PLSR models with reduced number of wavelengths (predictors) were built by selecting the variables with larger weights on the regression coefficients. Coefficient of prediction (R2p) and root mean square errors of prediction (RMSEP) obtained were 0.95 and 2.12; 0.97 and 1.68; 0.87 and 1.74, for potato starch, acacia gum and annatto, respectively. In conclusion, results showed that NIR spectroscopy is a useful screening technique for identification and quantification of adulteration in paprika.

中文翻译：

---

便携式近红外光谱仪用于快速鉴定掺假辣椒粉

摘要 辣椒粉是一种广泛食用的香料，使其成为不易检测的掺假目标。在本研究中，便携式近红外 (NIR) 光谱仪用于快速检测辣椒粉掺假。来自 5 个供应商的 9 个辣椒粉样品掺入了不同浓度的马铃薯淀粉、金合欢树胶和胭脂树胶（按重量计 0-36% 的马铃薯淀粉和金合欢胶，以及按重量计 0-18% 的胭脂树胶）。每种混合物 (n = 315) 的 NIR 光谱用作预测因子，通过偏最小二乘判别分析 (PLS-DA) 和偏最小二乘回归 (PLSR) 确定掺假。首先，PLS-DA 用于区分掺假和非掺假样品，以及掺假物的类型。这个方法被证明是有效的，对于构建的所有模型，特异性大于 90%，错误率小于 2%。PLSR 用于预测辣椒粉样品中掺杂物的浓度。此外，通过选择回归系数权重较大的变量来构建波长数量减少（预测变量）的 PLSR 模型。得到的预测系数（R2p）和预测均方根误差（RMSEP）分别为0.95和2.12；0.97 和 1.68；0.87 和 1.74，分别为马铃薯淀粉、阿拉伯胶和胭脂树。总之，结果表明 NIR 光谱是一种有用的筛选技术，可用于鉴定和量化辣椒粉中的掺假。通过选择对回归系数具有较大权重的变量来构建具有减少的波长（预测变量）数量的 PLSR 模型。得到的预测系数（R2p）和预测均方根误差（RMSEP）分别为0.95和2.12；0.97 和 1.68；0.87 和 1.74，分别为马铃薯淀粉、阿拉伯胶和胭脂树。总之，结果表明 NIR 光谱是一种有用的筛选技术，可用于鉴定和量化辣椒粉中的掺假。通过选择对回归系数具有较大权重的变量来构建具有减少的波长（预测变量）数量的 PLSR 模型。得到的预测系数（R2p）和预测均方根误差（RMSEP）分别为0.95和2.12；0.97 和 1.68；0.87 和 1.74，分别为马铃薯淀粉、阿拉伯胶和胭脂树。总之，结果表明 NIR 光谱是一种有用的筛选技术，可用于鉴定和量化辣椒粉中的掺假。