In this paper,Fourier transform infrared spectroscopy(FTIR),attenuated total reflection Fourier transform infrared spectroscopy(ATR-FTIR) and ultraviolet(UV) spectra of the samples from six different areas in Yunnan and Guizhou provinces were collected,and the ATR-FTIR data were processed by ATR calibration.The models for single spectra and low level data fusion were established using different pretreatments including first derivatives(FD),second derivatives(SD) and standard normal variables(SNV) combined with partial least squares discriminant analysis(PLS-DA) and support vector machine(SVM).Results showed that the corrected ATR-FTIR(ATR-FTIR-A) spectral information was in agreement with the FTIR spectra.Additionally,second derivative processing could improve the UV resolution.ATR-FTIR-A and ATR-FTIR-A-UV low level data fusion models had the highest accurate rate(100.00%) and the best parameter values,which were established to identify Ppolyphylla from different regions by PLS-DA and SVM,respectively.Therefore,ATR-FTIR-A,with the simple and nondestructive characteristic in sample pretreatment,had a better performance than FTIR in identification of Ppolyphylla of different geographical origins .It was accurate in practical production and application.However,ATR-FTIR-A-UV had a better performance than PLS-DA or SVM with only one kind of spectra.