论文部分内容阅读
目的:利用近红外特征谱段相关系数法建立定性模型对真伪龙齿进行快速鉴别。方法:在传统性状鉴别的基础上,使用近红外光谱仪的光纤附件采集光谱,以正品龙齿的近红外光谱为参照光谱,选择特征谱段,设定合适阈值,建立真伪龙齿定性鉴别模型,通过模型快速判断待测龙齿样品的真伪。结果:选定5 000~4 200 cm-1谱段为特征谱段,计算训练集样品中真伪龙齿的相关系数,设定阈值为92.67%,用10批验证集样品对模型进行验证,预测正确率为90%。结论:利用近红外特征谱段相关系数法建立的定性鉴别模型具有较好的预测能力,可以做为快速鉴别真伪龙齿的方法。
OBJECTIVE: To establish a qualitative model based on the correlation coefficient of near-infrared spectral bands for rapid identification of true and false dragon teeth. Methods: Based on the traits of traditional traits, the optical spectra of near-infrared spectrometer were collected, and the near-infrared spectrum of genuine dragon teeth was taken as the reference spectrum. The characteristic spectrum was selected and the appropriate threshold was set. , Through the model to quickly determine the test tooth sample authenticity. Results: The spectral bands between 5 000 and 4 200 cm-1 were selected as the characteristic spectral bands to calculate the correlation coefficient of the true and false teeth in the training set samples. The threshold value was set at 92.67%. The model was validated with 10 batches of validation set samples, The forecast accuracy is 90%. Conclusion: The qualitative discriminant model established by the correlation coefficient method of near-infrared feature spectrum has good predictive ability and can be used as a method to quickly identify authentic dragon teeth.