Summary: | 碩士 === 國立屏東科技大學 === 機械工程系 === 94 === Fluorescence image demonstrated its feasibility in classification of Aflatoxin of A. flavus from non Aflatoxin of A. niger and A. oryzae. When U band was used and the classification accuracy was 82.67% with 18 features(X1-X18) in Linear discriminate analysis. The correlation coefficient R2 was 0.85 in standard calibration curve when two parameter was used by Veratox ELISA kits and R2 was 0.98 when four parameter non linear model was applied by Myco. In Visible/NIR spectra calibration, results showed that SEP=0.017, =0.937, R.E.=11.49% and 88.51% accuracy classification could be obtained by MLR model for raw data of ELISA. In terms of prediction concentration of ELISA kits, PLSR was the fittest which had SEP=8.875, =0.910, R.E.=7.04% and 94.73% classification accuracy.
|