Abstract:
A technique for predicting the K-value of fish freshness using Fourier Transform-Near-Infrared (FT-NIR) coupled with chemometric algorithms was developed. FT-NIR spectra of 150 fresh fish were acquired, and their respective K-value estimated using HPLC procedures. Acquired spectra were pretreated using standard normal variate (SNV) to eliminate extraneous information. The developed FT-NIR coupled chemometric algorithm attained K-values of the correlation coefficient between 0.9471–0.9786 and the predictive deviation of the residuals (RPD) of 3.53–4.19. The performance of the ant colony partial least squares (ACO-PLS) compared to the other chemometric algorithms proved superior for the estimation of K-values of fish freshness with a correlation coefficient of 98.27 % for the training set and 97.86 % for the test set recorded. This implies that FT-NIR, coupled with different chemometric algorithms, has the potential to be deployed for accurate prediction of K-value as a freshness indicator.