學生姓名:
廖頌賢
指導教授:
蕭心怡
學期:
109下
摘 要:
The quality of aquatic products has been widely concerned by consumers, especially the freshness. Thus, the goods are generally frozen or refrigerated after purchase as soon as possible to maintain the freshness. However, customers have no insights on whether the products are fresh or frozen-thawed before purchasing due to some dishonest business tricks of selling frozen-thawed aquatic products as fresh to gain more profits. The purpose of this study was to predict the contents of TVB-N in aquatics using hyperspectral image technology (wavelength range from 400 to 1000nm) and to determine the products status and its storage time stated as fresh and freeze-thawed. The results showed that the prediction model (BP-ANN) based on the MSC corrected spectral data could accurately predict the TVB-N contents in oysters during storage, and its R2 p was 0.9721. In addition, PLS model had been established by using the spectral data of four different types of fishes to predict storage time while PLS-DA model was used to classify between fresh and frozen-thawed states. The best PLS models with Rcv 2 were obtained from the spectrum of the whole scale fish, 0.80 and 0.84 respectively, and the PLS-DA model with the accuracy of 100% was obtained from the spectrum of flesh side of fish fillets. In conclusion, hyperspectral image technology is an useful indicator to predict freshness and to distinguish between different status of aquatic products.
專題討論摘要_廖頌賢.pdf