基于光谱技术的支持向量机判别牛肉新鲜度

Assessment of beef freshness based on spectral technology with support vector machine

中文关键词:  可见/近红外光谱  支持向量机  牛肉新鲜度  无损分类

英文关键词:visible/near-infrared spectrum  support vector machine  beef freshness  non-destructive classification

中文摘要:

      目的 实现生鲜牛肉新鲜度等级的无损快速判别。方法 用可见/近红外光谱检测系统, 获取储存1~18 d的36块牛肉样品的400~1600 nm范围的光谱信息, 以挥发性盐基氮理化值为分类依据。用多元散射校正(MSC)、变量标准化(SNV)、SG平滑预处理方法处理光谱数据, 分别建立牛肉新鲜度的支持向量机分类模型。结果 MSC+SG预处理后所建立的分类模型预测能力最好, 训练集和测试集的回判识别率和预测识别率分别为96.30%、100%, 验证集的识别率为88.89%。结论 可见/近红外光谱结合支持向量机, 对牛肉新鲜度进行无损快速判别是可行的。

英文摘要:

      Objective To establish a rapid and non-destructive discrimination method of freshness level of fresh beef. Methods A total of 36 beef samples’ reflectance spectra in 400~1600 nm were collected by a laboratory visible/near-infrared spectroscopy system. These samples were stored for 1~18 d. The total volatile basic nitrogen (TVB-N) values of sample were taken as the reference of classification. The multiplicative scatter correction (MSC), standard normalized variate (SNV), and savitzky-golay (SG) smoothing methods were used as the pretreatment method for reflectance spectra processing, respectively. The classifier models of support vector machine (SVM) were built for identification of beef freshness. Results The MSC combined with SG smoothing was the best pretreatment, and the predictive ability of classification model built was the best. The identification rates of training set and prediction set were 96.30% and 100%, respectively. The identification rate of validation set was 88.89%. Conclusion It is feasible to use visible/near-infrared spectrum combined SVM for rapid and non-destructive identification of freshness level of fresh beef.