论文部分内容阅读
目的探索基于FTIR离散平稳小波变换结合支持向量机(support vector machine,SVM)分类法的中药紫花地丁的质量控制新模式。方法采用衰减全反射傅里叶变换红外光谱法直接快速测定中药紫花地丁与同属植物多花堇菜和戟叶堇菜的FTIR,运用基于离散平稳小波变换进行特征向量的提取,通过分析比较后选取第4、5层分解层的特征向量用于支持向量机的训练与验证。结果通过对不同产地的90个样本的验证,紫花地丁与同属植物多花堇菜和戟叶堇菜的识别率达100%。结论基于FTIR离散平稳小波变换结合支持向量机分类法的中药紫花地丁与同属植物多花堇菜和戟叶堇菜的分类鉴别方法具有非常好的效果。
Objective To explore a new quality control mode of Viola yedoensis based on FTIR discrete stationary wavelet transform combined with support vector machine (SVM) classification. Methods FTIR was directly and rapidly determined by attenuated total reflectance Fourier transform infrared spectroscopy (FTIR). The FTIR of the genus Violata and the genus Violata sibirica was determined by the FTIR method. The eigenvectors were extracted by using the discrete stationary wavelet transform. The eigenvectors of layer 4 and layer 5 are selected for the training and verification of support vector machines. Results Through the validation of 90 samples from different areas, the identification rate of Viola yedoensis with the same genus Viola and violaceum reached 100%. Conclusion The method based on FTIR discrete stationary wavelet transform combined with support vector machine classification of Viola yedoensis and the same genus Viola and Viola halophila classification and identification method has a very good effect.