Abstract:  A nondestructive measurement model for predicting soluble solids contents and total acidity contents of yellow sweet peppers by near infrared spectroscopy and improved partial least squares was designed. Improved partial least squares （MPLS） calibration model of soluble solids contents and total acidity contents of sweet peppers were established respectively in two wavelength ranges of 400-2 500 nm and 400-980 nm+1 108-1 900 nm， then the optimal model was used to predict. The results showed that standard normal variate （SNV） model obtained the best result for soluble solids contents （SSC） prediction with a correlation of determination of calibration （RCV） of 0926 9 and standard error of cross validation （SECV） of 0186 2， and a correlation coefficient of prediction （RP） of 0924 8 and standard error of prediction （SEP） of 0158 9. None model obtained the best result for total acidity contents prediction with a correlation of determination of calibration （RC） of 0868 0 and standard error of cross validation （SECV） of 0012 5， and a correlation coefficient of prediction （RP） of 0903 8 and standard error of prediction（SEP）of 0011 1. The results indicated that it was feasible for determination of soluble solids contents and total acidity contents of sweet pepper based on near\|infrared diffuse reflectance spectroscopy and improved partial least squares， and the soluble solids and total acidity contents in sweet pepper had a significant correlation with the near infrared spectroscopy.

Key words:  , near infrared spectrum；improved partial least squares；sweet pepper；soluble solids content；total acidity content