基于智能视觉技术的白及饮片真伪快速辨识方法研究

目的 利用电子眼和色差仪分别对白及Bletillastriata及其伪品进行辨识，旨在建立白及真伪的快速辨识方法。方法 首先基于《中国药典》、地方标准和HPLC指纹图谱对多采集来源的134批白及及其伪品样品进行综合鉴别，以确定标杆辨识信息（Y）。然后基于电子眼和色差仪分别获取上述样品的智能视觉信息（X）。最后利用Matlab建立Y=F(X)模型，即分别建立白及真伪二分类（白及、非白及）和四分类（白及、天麻Gastrodiaelata、玉竹Polygonatumodoratum、黄花白及Bletilla ochracea）的主成分分析-判别分析（principal component analysis-discriminant analysis,PCA-DA）、偏最小二乘-判别分析（partial least squares-discriminant analysis,PLS-DA）、最小二乘-支持向量机（least squares-support vector machine,LS-SVM）和K最近邻（K nearest neighbor,KNN）辨识模型并验证。结果 经留一法交互验证...

Authenticity rapid identification method of Bletilla striata decoction pieces based on intelligent vision technology

Objective To identify Baiji (Bletilla striata) and its approximate decoction pieces by electronic eye and colorimeter, so as to establish a rapid identification method for authenticity of B. striata decoction pieces. Methods Firstly, 134 batches of B. striata and its counterfeit samples from multiple sources were comprehensively identified based on Chinese Pharmacopoeia, local standards and HPLC fingerprint to determine the benchmarking information (Y). Then intelligent visual information (X) of the above samples was obtained based on electronic eye and colorimeter. Finally, the Y = F(X) model was established by Matlab, that is, the principal component analysis-discriminant analysis (PCA-DA), partial least squares-discriminant analysis (PLS-DA), least squares support vector machine (LS-SVM) and K nearest neighbor (KNN) of 134 batches of samples were established and verified for the Authenticity binary classification (B. striata, non-Bletilla striata) and the four classification B. striata, Tianma (Gastrodia elata), Yuzhu (Polygonatum odoratum) and Huanghuabaiji (Bletilla ochracea)]. Results After the leave-one-out interactive verification, the optimal identification model based on electronic eye is KNN model, and the positive rate is 99.25%; The four-class optimal identification model is the LS-SVM model, and the positive rate is 97.01%. The optimal identification model based on colorimeter is KNN model, and the positive rate is 99.25%; the optimal identification model of four classifications is PLS-DA model, and the positive rate is 97.67%. Based on two types of intelligent vision and multi-source information fusion technology, the optimal identification model of two classifications is PCA-DA and KNN model, and the positive rate is 98.51%, which is lower than that before fusion; The optimal identification model of four classifications is LS-SVM model, and the positive rate is 97.76%, which is higher than that before fusion. Conclusion Two kinds of intelligent vision technology can be used for rapid identification of authenticity of B. striata decoction pieces. The establishment of this method can provide reference for the rapid and accurate identification of the quality of traditional Chinese medicine.