利用可见/近红外(400~1 000 nm)及近红外(900~1 700 nm)高光谱成像技术结合特征波长筛选方法对安格斯牛、力木赞牛、西门塔尔牛3个品种的牛肉进行鉴别研究，且测定肉样的色泽、嫩度、pH值以及水分、脂肪、蛋白质含量。根据不同波段光谱的特点，分别对原始光谱进行预处理，并利用SPA、IRF和IRF-SPA方法筛选特征波长，建立基于全波段及特征波长下的PLS-DA牛肉品种鉴别模型。结果显示：400~1 000 nm波段采用SNV-IRF-SPA-PLS-DA方法建立的模型最优，校正集与预测集准确率分别为98.56%和97.12%，900~1 700 nm波段采用SG-SPA-PLS-DA方法建立的模型准确率为94.09%和96.04%，说明不同波段高光谱对牛肉品种识别均有较好的效果；400~1 000 nm波段的识别准确率优于900~1 700 nm，说明3种牛肉在色泽纹理上的差异比成分含量显著。研究表明，利用高光谱成像技术结合特征波长筛选方法能够获得较好的牛肉品种鉴别效果。

This paper focused on the research on identifying and classifying for beef varieties of Angus, Limousin and Simmental, by using visible/near-infrared(400~1 000 nm) and near infrared(900~1 700 nm) hyperspectral technologies combined with different characteristic wavelengths selection methods. Meanwhile, the contents of color, tenderness, pH value, moisture, fat and protein were measared. Pretreatment methods were used to process original spectrum respectively according to the characteristics of different spectrum bands; the characteristic wavelengths were extracted by using SPA, IRF and IRF-SPA; then PLS-DA model was applied to identify the different beef varieties under characteristic wavelengths and fullwave bands. Results showed that SNV-IRF-SPA-PLS-DA models achieved the optimal performance in 400~1 000 nm, and the accuracy of the correction set and prediction set was 98.56% and 97.12%, respectively. SG-SPA-PLS-DA models achieved the optimal performance in 900~1 700 nm, and the accuracy of the correction set and prediction set was 94.09% and 96.04%, respectively. There were good effects for beef varieties identification in different hyperspectral bands. The identification accuracy in 400~1 000 nm bands was better than in 900~1 700 nm bands, which explained that the differences of color and texture were more significant than the component contents among the 3 varieties beef. The research indicated that combined hyperspectral technologies with characteristic wavelengths selection methods can obtain a better recognition effect of beef varieties.