Clustering Analysis of Near-Infrared Spectra of Rhubarb after Wavelet Transform

【摘要】：[Objective] To conduct map scanning of 41 types of rhubarb by near-infrared diffuse reflectance spectrometry, and to provide a new method for the pharmacognostic identification of rhubarb. [Method] Near-infrared spectra were compressed by wavelet transform technique. Training time was reduced; rhubarb samples in different production areas were identified by both near-infrared spectroscopy and clustering analysis method. [Result] After compression of wavelet transform by near-infrared spectroscopy, spectral variables reduced from 700 to 44. Wavelet compressed near-infrared spectra were inputted to clustering tree identification model. Based on clustering analysis, certified rhubarb was identified. And the recognition correct rate reached 82.93%. [Conclusion] Near-infrared diffuse reflectance spectrometry was a rapid and simple identification and analysis technology, and could be used for the quality control of rhubarb.