On-line prediction of pH values in fresh pork using visible/near-infrared spectroscopy with wavelet de-noising and variable selection methods
2012
Liao, Yitao | Fan, Yuxia | Cheng, Fang
The visible/near-infrared (Vis/NIR) reflectance spectroscopy as an on-line approach to assess the pH value in fresh pork was investigated. Multivariate calibration was carried out by using chemometrics. Discrete wavelet transform was applied to de-noise the spectra scanned on-line, and several variable selection methods were proposed to simplify the calibration models. The study found that the model based on the spectra de-noised by Daubechies 6 wavelet (db6) at decomposition level 6, soft thresholding strategy and minimaxi threshold estimator gave reasonable performance (r>0.900, root mean square error of calibration (RMSEC)=0.100, cross validation (RMSECV)=0.139 and prediction (RMSEP)=0.125). Then, only 15% variables from this model were selected via the method of uninformative variable elimination to develop a simpler model, of which the performance deterioration could be ignored. The results showed that Vis/NIR can be used to predict pH value in fresh pork on-line, and variable selection can provide a simpler, more cost-effective calibration model.
Show more [+] Less [-]AGROVOC Keywords
Bibliographic information
This bibliographic record has been provided by National Agricultural Library