论文部分内容阅读
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。
Objective To establish Jinhua ham quality grade evaluation model. Methods The samples of Jinhua ham with different quality grades were detected by hyperspectral imager. Principal component analysis (PCA) and Partial Least Squares (PLS) were used to analyze the obtained image information with data analysis software. Results The PCA and PC2 contribution rates were 86% and 11%, respectively, with a total contribution rate of 97%. In the discriminant model established by PLS, the overall recognition coincidence rate of training set and verification set is 96.19% and 89.52% respectively. Conclusion The combination of hyperspectral imaging with a certain pattern recognition method to establish the evaluation model is a feasible Jinhua ham quality grade test new technology.