基于可见/近红外光谱的鹅鸭混合绒定量检测研究

鹅绒和鸭绒的外观相似但在品质上鹅绒优于鸭绒,各国羽绒毛标准对鹅绒毛中的鸭绒毛含量都有最高限定。传统检测方法为高倍显微镜目测法,该方法劳动强度大,且不适宜大批量样本的分析及现场快速检测。利用可见/近红外光谱结合连续投影算法(SPA)特征波长选择的建模方法对鹅绒中混有鸭绒含量进行了定量检测。在450～930nm范围内,通过SPA选择的8个特征波长建立多元线性回归模型,取得了较好的预测结果,相关系数为0.983,校正均方根误差(RMSEC)为5.44%,预测均方根误差(RMSEP)为5.75%,有望用于羽绒毛品质的快速检测。

Quantitative analysis of goose and duck mixed down using visible/NIR spectroscopy

Goose down and duck down have very similar appearance but the quality of goose down is better than that of duck down in general. There is a highest allowable limit as specified by the various national standards of feather and down for the percentage of duck feather or down mixed in goose feather or down. Traditional detection method, manual inspection with a high-scale microscope, is labor intensive and not suitable for large-volume samples analysis and on-site rapid testing. In the present paper, visible/near-infrared (NIR) spectroscopy combined with successive projection algorithm (SPA) for characteristic wavelengths selection was used to determinate the content of duck down mixed in goose down. In the range of 450-930 nm, the multiple linear regression (MLR) model established with the 8 characteristic wavelengths selected by SPA achieved good prediction, the correlation coefficient of 0.983, root mean square error of calibration (RMSEC) of 5.44%, and root mean square error of prediction (RMSEP) of 5.75%. Therefore, it is expected to be used for rapid detection of feather and down quality in future.