Study On The Secondary Feature Extraction Method Applied To The Tea Origin Identification

Chinese is the hometown of tea and it is rich diversified tea. Tea has importanthealth care functions for improving human immunity and maintain beauty and keepyoung etc. Thus, it is important for both manufacturers and consumers toidentification the tea quickly and accurately. The origin of tea is studied by Fourierinfrared spectrum and spectral characteristics basis theory. The main work of this taskincludes:1. Spectral data of new tea samples are collected and spectral characteristicsbasis is established by selecting the appropriate sample. The spectral data acquisitionof new tea project on the spectral characteristics basis that can reflect the origin of tea.This can be used to verify feature characteristic factor L1, the second feature factorL2and the third feature factor L3of the biggest information difference.The feasibilityand effectiveness of the characteristics basis can be verified.2. According to the different of regional characteristics of tea origin, thesecondary feature extraction method was proposed on the basis of the principle of theprincipal component analysis This method can be applied for identifying anddistinguishing of infrared spectroscopy of characteristics of tea varieties growing atdifferent tea origin. At the same time, the band selection, pretreatment method,content was studied in the process of research.3. To participate in the regional tea producing area, the number of the foundationof tea producing area were analyzed, and the optimal basis is determined.25kinds oftea and16kinds of tea are respectively identified with secondary feature extractionmethod. The results show that the mass organizations of tea in the same region isgood and the tea in the same region distribute in the same area.The secondary featureextraction method is applied to study the tea origin and its feasibility of theapplication in these aspects is validated.4.Six categories of tea are identified by the secondary feature extraction method.The effect of tea participation in infrastructure of different kinds of tea are compared and analyzed, we confirm the participation in infrastructure of green tea, yellow tea,white tea, green tea, red tea and black tea. After the secondary feature extraction themain functions of different kinds of feature factors in participating in infrastructureare analyzed, we have selected the suitable feature factors as characteristic foundation;we merge the feature factors with similar functions and build characteristic foundationwith feature factors with different function, thus we can identify six categories of teaeffectively. Establish discriminant, identification of the six categories of teadiscriminant rate of discriminant are verified, get the correct recognition rate of80%.