Search

In the present paper, the authors extracted active components of herba epimedii and their important features using Fourier transform infrared spectroscopy (FTIR), correlation coefficient comparison, and multilevel wavelet analysis. The extracted features were then used to classify herba epimedii via radial basis function (RBF) neural network. There were altogether 250 samples of the medicine with various different types, including epimedium brevicornu Maxim., E. sagittatum (Sieb. et Zucc. ) Maxim, E. pubescens Maxim., E. koreanum Nakai and E. wushanense T. S. Ying. An important component of herba epimedii, herba epimedii icariin, has a special peak at 1 259 cm(-1) on the FTIR spectra obtained from the methanol extraction, which is consistent with the result obtained by traditional HPLC qualitative analysis. Therefore, this special peak can be used to determine if herba epimedii contains herba epimedii icariin. Furthermore, large variations in the spectrum caused by low content of icariin, weak absorption peaks and noise were successfully removed by applying correlation coefficient comparison and multilevel wavelet analysis, which significantly increased the quality of classification of RBF neural network. This paper creates a framework of fast identification of herba epimedii icariin in raw herba epimedii by FTIR spectra via wavelet analysis and RBF neural network.

Infrared spectra of Papaver somniferum L. and Papaver rhoeas were obtained directly, quickly and accurately by Fourier transform infrared spectroscopy (FTIR)with OMNI sampler. Discrete stationary wavelet transform was used to extrude local region of infrared spectra of Papaver somniferum L. and Papaver rhoeas. The difference of infrared spectra between Papaver somniferum L. and Papaver rhoeas was extruded. Accurate identification rate is improved greatly. One dimensional discrete stationary wavelet transform was implemented to the infrared spectra of Papaver somniferum L. and Papaver rhoeas. The difference between Papaver somniferum L. and Papaver rhoeas was observed at all scales in wavelet domain. Two scales, at which the difference between Papaver somniferum L and Papaver rhoeas is the most obvious, were selected to extract the features of Papaver somniferum L. and Papaver rhoeas. A feature vector including eight feature parameters was constructed. The feature vector was input to RBFNN for training in order to accurately identify Papaver somniferum L. and Papaver rhoeas. In experiment, the authors used one hundred and twenty-eight couples of data of Papaver somniferum L. and Papaver rhoeas (including seventy-eight couples of training samples and fifty couples of testing samples). The experimental results show that it is effective to apply discrete stationary wavelet transform on the basis of FTIR to identify the Papaver somniferum L. and Papaver rhoeas. The accurate identification rate of Papaver somniferum L and Papaver rhoeas is 99.8% and 99.9% respectively.

An improved genetic algorithm was used to implement an automated wavelength selection procedure for use in building multivariate calibration models based on partial least squares regression (PLS). The region selecting by genetic algorithms (R-SGA) was applied in building calibration model of soluble solid content (SSC) of Western pear, and the numbers of latent variables used to build calibration model were further reduced. The Fourier transform near infrared reflectance (FT-NIR) spectra were processed by GA after MSC or SNV, and four PLS calibration models were built by using the optimal combinations of these sub-regions. Meanwhile, the full region selecting PLS (Fr-PLS) models were developed. The R-SGA models variables were 434, 496, 310 and 496, for Early Red Comice, Wujiuxiang, Cascade and Kang Buddha, respectively. Despite the complexity of the spectral data, the R-SGA procedure was found to perform well (RMSEP = 0.428, 0.567 for Early Red Comice and Kang Buddha, respectively), leading to calibration models that significantly outperform those based on full-spectrum analyses (RM-SEP = 0.518, 0.633). The prediction precision of GA-PLS models was similar to that of Fr-PLS for Wujiuxiang and Cascade, with RMSEP of 0.696/0.694 and 0.425/0.421 respectively. This work proved that the R-SGA could find optimal values for several disparate variables associated with the calibration model and that the PLS procedure could be integrated into the objective function driving the optimization.