基于高光谱成像技术的金华火腿无损分级检测研究

目的 建立金华火腿的质量等级评判模型。 方法 采用高光谱成像仪检测不同质量等级的金华火腿样本, 结合数据分析软件对得到的图像信息作主成分分析(PCA)和偏最小二乘(PLS)分析。 结果 用PCA处理, 第一主成分(PC1)和第二主成分(PC2)的贡献率分别为86%和11%, 总贡献率为97%。PLS建立的判别模型中, 训练集和验证集的总体识别吻合率分别为96.19%和89.52%。 结论 将高光谱成像技术与一定的模式识别方法相结合建立评判模型, 是一种可行的金华火腿质量等级检验新技术。

Rapid non-destructive detection of grade classification in Jinhua ham by hyperspectral imaging technique

Objective To establish an available method to evaluate the quality grades of Jinhua ham. Methods The samples of variety grades of Jihua ham were detected by hyperspectral imaging system, and the collected imaging informations were analyzed by using Principal Component Analysis (PCA) and Partial Least Squares (PLS) with data software. Results The PCA analysis results showed that the variances of PC1 and PC2 reached 86% and 11%, respectively, the total variance reached 98%. The detection model constructed with PCA showed that the total accuracy prediction of training set and prediction set for grade classification of Jinhua ham reached 96.19% and 89.52%, respectively. Conclusion Construction of detection model combined with hyperspectral imaging technique and suitable recognized methods is one of new practical techniques to detect Jinhua ham grade without destruction.