Determination of sweet corn seed vigor by near infrared spectra and chemometrics

【Objective】To develop a new method to measure sweet corn seed vigor by near infrared(NIR) spectroscopy and chemometrics, so as to provide a new method for seed screening in the seed industry.【Method】Near infrared spectra of sweet corn seed under reflection and transmission modes were recorded. Then outliers were identified and eliminated by principal component analysis(PCA) and Monte Carlo cross validation methods. The proper preprocessing methods and variables selection methods were applied to establish and select the partial least squares(PLS) prediction model.【Result】For the vitality index quantitative analysis model of diffuse reflection, a total of 532 samples were used for modeling study. The best preprocessing method was Savitzky-Golay derivative(SG) +Mean Center(MC), and the optimal variable selection method was competitive adaptive reweighted sampling(CARS). The correction correlation coefficient(R_c), cross validation correlation coefficient(R_cv), prediction correlation coefficient(R_p), root mean square error of correction(RMSEC), root mean square error of cross validation(RMSECV) and root mean square error of prediction(RMSEP) of the model were 0.826, 0.783, 0.663, 0.137, 0.151 and 0.199, respectively. For the vitality index quantitative analysis model of transmission, a total of 415 samples were used for modeling study. And the best preprocessing method was SG first derivative smoothing method, and the optimal variable selection method was correlation coefficients(CC). The model performance parameters R_c, R_cv, R_p, RMSEC, RMSECV and RMSEP were 0.783, 0.680, 0.728, 0.121, 0.142 and 0.133, respectively, there was no over-fitting in this model, indicating that the transmission model of spectral acquisition might be more suitable for the determination of seed vigor index.【Conclusion】More information about sweet corn seed vigor can be obtained through transmission spectrum, and transmission module is a better method for spectral acquisition and prediction of seed vigor.