Abstract:

In order to achieve rapid nondestructive determination of moisture and protein contents in Alaska pollock surimi, near-infrared quantitative models were established by fitting the near-infrared diffuse reflectance spectral data of Alaska pollock surimi with its chemical values. The models were further developed by partial least square (PLS) regression and the correlation coefficients of the calibration models obtained were above 0.96 and 0.98, respectively, which gave more reliable prediction than its near-infrared transmission counterpart. The RSD values for 10 replicate predictions of moisture and protein contents in Alaska pollock surimi were both less than10% and the relative percentage difference (RPD) values were above 3, showing excellent effectiveness of the models developed in this study in predicting moisture and protein contents in Alaska pollock surimi.

Key words: surimi, NIRS, moisture, protein, non-destructive determination