Discrimination of varieties of Chinese bayberry using near infrared spectroscopy

A new method for discrimination of varieties of Chinese bayberry by means of infrared spectroscopy (NIRS) (325-1075nm) was developed. A relation has been established between the reflectance spectra and Chinese bayberry varieties. The dataset consist of a total of 69 samples of Chinese bayberry. First, the data was analyzed with principal component analysis. It appeared to provide the best clustering of the varieties of Chinese bayberry. PCA compressed thousands of spectral data into a small quantity of principal components and described the body of spectra; the scores of the first 20 principal components computed by PCA had been applied as inputs to a back propagation neural network with one hidden layer. 69 samples contained three varieties were selected randomly, then they were used to build BP-ANN model. This model had been used to predict the varieties of 15 unknown samples; the residual error for the calibration samples is 1.508458 x 10(-6). The recognition rate of 100% was achieved. The result achieved by using PCA-BP method is much better than the results achieved by only using the PCA method. This model is reliable and practicable. So this paper could offer a new approach to the fast discrimination of varieties of Chinese bayberry.