The model to measure the leaf area index(LAI）of wheat with visible/near-infrared reflectance spectra is improved.The wheat canopy reflectance spectra is pretreated by different methods,and then the LAI estimation models are established by partial least square (PLS）algorithm to comparative analysis different pretreatments.It is found that the pretreatment method of wavelet denoising combined with first derivative can eliminate the noise and background information of the original spectra most effectively,with the calibration R-square 0.849 and prediction R-square 0.835,respectively.For optimizing the model，the pretreated spectra are analyzed using principal component analysis (PCA),and the anterior 4 principal components,which accounted for 84.867% variation of the original spectral information,are used as the input variables to built the LAI estimation model by least square support vector regression(LS-SVR）algorithm.The calibration R-square and prediction R-square of LS-SVR model are 0.905 and 0.883,respectively,higher than that of PLS model,which indicates that the LS-SVR model is more accurate.The results suggest that it is feasible to improve the accuracy of the LAI estimation model by eliminating the soil background information of original spectra with the pretreatment method of wavelet denoising combined with first derivative,and the LS-SVR algorithm is a preferred method for model building.