Chemical spoilage extent traceability of two kinds of processed pork meats using one multispectral system developed by hyperspectral imaging combined with effective variable selection methods.

The feasibility of hyperspectral imaging (HSI) (400-1000nm) for tracing the chemical spoilage extent of the raw meat used for two kinds of processed meats was investigated. Calibration models established separately for salted and cooked meats using full wavebands showed good results with the determination coefficient in prediction (R2P) of 0.887 and 0.832, respectively. For simplifying the calibration models, two variable selection methods were used and compared. The results showed that genetic algorithm-partial least squares (GA-PLS) with as much continuous wavebands selected as possible always had better performance. The potential of HSI to develop one multispectral system for simultaneously tracing the chemical spoilage extent of the two kinds of processed meats was also studied. Good result with an R2P of 0.854 was obtained using GA-PLS as the dimension reduction method, which was thus used to visualize total volatile base nitrogen (TVB-N) contents corresponding to each pixel of the image.