Hits:

Indexed by:期刊论文

Date of Publication:2018-06-01

Journal:FOOD ANALYTICAL METHODS

Included Journals:SCIE

Volume:11

Issue:6

Page Number:1701-1710

ISSN No.:1936-9751

Key Words:Fish morphologies; Quality; Hyperspectral imaging; Rapid detection

Abstract:To investigate the effect of morphologies on assessment of fish quality, hyperspectral images were obtained from intact fish with scales, intact scaled fish, skin side of fish fillets, and flesh side of fish fillets, with wavelength range from 400 to 1000 nm. The storage time prediction and adulteration of fresh and frozen-thawed intact fish and fish fillets were studied, respectively. Different preprocessing methods were applied to process the original spectra. Partial least square (PLS) regression methods were used to predict the storage time of intact fish and fish fillets, with spectra obtained from the four fish morphologies mentioned above. Partial least square discriminant analysis (PLS-DA) algorithm was used to classify fresh and frozen-thawed intact fish and fish fillets with spectra obtained from the four morphologies. For storage time prediction of intact fish, the optimal PLS models were developed with spectra from intact scaled fish, with R-cv(2) of 0.80 and 0.84 for fresh and frozen-thawed fish, respectively. For storage time prediction of fish fillets, the best PLS models were developed with spectra measured from flesh side of fish fillets, with R-cv(2) of 0.84 and 0.84 for fresh and frozen-thawed fish fillets, respectively. The PLS-DA models developed with the spectra from intact scaled fish presented a better result with accuracy of 100%. The models developed with spectra obtained from intact fish with scales, flesh, and skin sides of fish fillets also demonstrated a good classification results with accuracy of 92.5, 97.5, and 97.5%.