Pork Grade Evaluation Using Hyperspectral Imaging Techniques

摘要

The method to evaluate the grade of the pork based on hyperspectral imaging techniques was studied. Principal component analysis (PCA) was performed on the hyperspectral image data to extract the principal components which were used as the inputs of the evaluation model. By comparing the different discriminating rates in the calibration set and the validation set under different information, the choice of the components can be optimized. Experimental results showed that the classification evaluation model was the optimal when the principal of component (PC) of spectra was 3, while the corresponding discriminating rate was 89.1 % in the calibration set and 84.9% in the validation set. It was also good when the PC of images was 9, while the corresponding discriminating rate was 97.2% in the calibration set and 91.1% in the validation set. The evaluation model based on both information of spectra and images was built, in which the corresponding PCs of spectra and images were used as the inputs. This model performed very well in grade classification evaluation, and the discriminating rates of calibration set and validation set were 99.5% and 92.7%, respectively, which were better than the two evaluation models based on single information of spectra or images.