为了研究快速识别轻微损伤壶瓶枣与完好壶瓶枣的有效方法，本文以轻微损伤壶瓶枣和完好壶瓶枣为研究对象，动态采集轻微损伤壶瓶枣和完好壶瓶枣的近红外光谱数据。采用S-G平滑与多元散射校正(MSC)相结合的方法预处理光谱数据，分别以预处理后的全光谱(FS)数据和采用主成分分析(PCA)法提取主成分、采用连续投影算法(SPA)提取特征波长作为输入变量，建立偏最小二乘判别分析(PLS-DA)和最小二乘支持向量机(LS-SVM)模型，比较4种损伤壶瓶枣及完好壶瓶枣的判别准确性。结果表明：采用PCA 提取主成分有较明显的优势，对4种损伤壶瓶枣的判别准确性均能满足实际要求，且采用PCA-LS-SVM模型对4种轻微损伤壶瓶枣和完好壶瓶枣的正确判别率最佳，分别达到100%、86%、100%、100%和100%，总的正确判别率为97.2%。该研究为轻微损伤壶瓶枣的动态判别提供了新的理论基础。

This study aimed to identify a rapid effective method to distinguish intact and subtly bruised Huping jujubes by the dynamic collection of their near infrared (NIR) spectral data. A combination of the Savitzky-Golay (S-G) and multiplicative scatter correction (MSC) methods were used for the preprocessing of spectral data. Full spectrum (FS) data obtained after preprocessing, major component data extracted by principal component analysis (PCA), and characteristic wavelength data extracted by successive projections algorithm (SPA) were used as input variables for the construction of models by partial least squares discriminant analysis (PLS-DA), or using the least squares-support vector machine (LS-SVM). The accuracy of these models in discriminating between the four types of intact and subtly bruised Huping jujubes was determined. The results of these analyses revealed the obvious advantages of PCA use for the extraction of the major components of Huping jujubes; in addition, this (PCA) data fulfilled all practical requirements for the accurate discrimination of all four types of subtly bruised samples. The PCA-LS-SVM model demonstrated optimal accuracy in the discrimination of four types of subtly bruised and intact Huping jujubes (100%, 86%, 100%, 100%, and 100%, respectively), , resulting in a total discrimination accuracy of 97.2%). In conclusion, this study provides a new theoretical basis for the dynamic discrimination of subtly bruised Huping Jujube.

国家自然科学基金资助项目(31271973)；高等学校博士学科点专项科研基金资助项目(20101403110003)；山西省自然科学基金资助项目(2012011030-3)