Categorization and authentication of Beijing-you chicken from four breeds of chickens using near-infrared hyperspectral imaging combined with chemometrics

摘要

Beijing-you chicken (BJY) is a classic chicken breed originating from the Beijing area. BJY is popular among the local people and faces the risk of counterfeiting in the market. In this study, hyperspectral imaging in tandem with multivariate analyses were employed to identify and authenticate BJY among four common chicken breeds. Images of chicken breasts (50 each) were collected in the near-infrared range (900-1,700 nm) and five pretreatment methods were implemented individually after spectra extraction. Different combinations of pretreatments and grouping methods were used with two statistical classifiers (k-nearest neighbor and support vector machine) before and after feature wavelength selection using successive projections algorithm. Finally, a comparatively satisfactory result was achieved, and the accuracies reached 100% for the calibration set and 92% for the prediction set, in which BJY could almost be identified with a 98% accuracy rate. The results indicated that the hyperspectral imaging technique was a promising method for industrial applications to address fraud in determining chicken breeds. Practical Applications Beijing-you chicken is a precious breed of chicken in China that has been applied for geographical indication products. The protection of this variety can provide theoretical basis and technical support for the division of geographical indications. Therefore, a fast and effective identification method is needed for the classification of Beijing-you chicken breed. Hyperspectral imaging is a popular method for varieties classification. In our study, the satisfying results revealed that this method could be used for classification of Beijing-you chicken when necessary. The results will be useful for the application of hyperspectral imaging technology in the classification of Beijing-you chicken. Moreover, this method has great potential for real-time online detection in factories for future work.