In this thesis, it took the non-destructive rapid testing of solid soluble content (SSC) in tomatoes as example. The near-infrared (NIR) spectra data of big and small tomatoes were collected by linear variable filter (LVF) NIR spectrometer and digital light processing (DLP) NIR spectrometer respectively. The average NIR spectra of big and small tomatoes and the difference spectra were calculated for LVF and DLP spectra respectively. The characteristics of the NIR spectra data of the two types of tomatoes collected by LVF and DLP spectrometer were compared respectively. Principal component analysis (PCA) was done on the LVF and DLP spectra respectively, and the distribution of the scores of the first 3 principal components were compared. The data were divided into calibration and external validation sets according to the SSC gradient. Partial least squares regression combined with a full cross-validation algorithm was applied to develop the quantitative calibration models of SSC in tomato for the spectra data collected by LVF and DLP spectrometer respectively. It is demonstrated by the result that: (1) The spectral characteristics of the average spectra and difference spectra of LVF-NIR spectra of big and small tomatoes are similar to those of DLP-NIR spectra, which indicates that it is feasible to carry out non-destructive and rapid testing of SSC in tomato by the LVF and DLP NIR spectrometers. (2) The separation trend of the score scatters of the first 3 principal components of LVF-NIR spectral data of big and small tomatoes was not obvious, while there is little separation trend for that of DLP-NIR spectral data. (3) The ratio performance deviation (RPD) values of the models developed by the LVF-NIR spectral data were no less than 2.11. Among them, the preprocessing of normalization acquired the optimized model, of which the number of factors (Nf), determination of calibration (R2C), root mean square error of calibration (RMSEC), determination of cross validation (R2CV), root mean square error of cross validation (RMSECV), RPD, correlation coefficient of prediction (rP) and root mean square error of prediction (RMSEP) were 8, 0.949 1, 0.27, 0.899 9, 0.38, 3.16, 0.882 6 and 0.63 respectively. The RPD values of the models developed by the DLP-NIR spectral data were no less than 1.60. Among them, the preprocessing of normalization acquired the optimized model, of which the Nf, R2C, RMSEC, R2CV, RMSECV, RPD, RP and RMSEP were 5, 0.823 5, 0.49, 0.728 6, 0.62, 1.94, 0.788 4 and 0.80 respectively. This thesis will, to some extent, provide reference to the non-destructive and rapid testing of SSC in tomatoes and the selection and evaluation of the non-destructive and rapid instrument for testing the quality of fruits and vegetables.