基于高光谱图像技术的冷却羊肉颜色检测

s for the current problems of contact and destruction existed in the processing of chilled lamb color detection, the hyperspectral imaging technique was used for detecting the parameters of vacuum-packed lamb at the chilled storage, which are lightness (L*), redness (a*), yellowness (b*) and color (e*). A total number of 140 representative samples of lamb spine were collected as the test material in the study, their hyperspectral imaging information was obtained, and the color parameters (L*, a*, b*and e*) of these samples were also determined. After spectral extraction and preprocessing, partial least squares regression (PLSR) models of all color parameters were established on the whole bands and the characteristic bands obtained from genetic algorithms (GA) separately. The results showed PLSR models established on GA characteristic bands were better than those on the whole bands. As for L*, a*, b* and e*, the correlation coefficients of calibration set (RC) were 0.98, 0.92, 0.83 and 0.93, and root mean square errors of calibration set (RMSEC) were 0.87, 0.61, 0.97 and 0.11, respectively. The correlation coefficients of prediction set (RP) were 0.98, 0.90, 0.77 and 0.87, and root mean square errors of prediction set (RMSEP) were 0.93, 0.67, 1.10 and 0.15, respectively. The hyperspectral imaging technique can quickly and nondestructively detect the color parameters of vacuum-packed chilled lamb at different storage time.