Hongbin Pu

Classification of fresh and frozen-thawed pork muscles using visible and near infrared hyperspectral imaging and textural analysis

The potential of visible and near infrared hyperspectral imaging was investigated as a rapid and nondestructive technique for classifying fresh and frozen-thawed meats by integrating critical spectral and image features extracted from hyperspectral images in the region of 400-1000 nm. Six feature wavelengths (400, 446, 477, 516, 592 and 686 nm) were identified using uninformative variable elimination and successive projections algorithm. Image textural features of the principal component images from hyperspectral images were obtained using histogram statistics (HS), gray level co-occurrence matrix (GLCM) and gray level-gradient co-occurrence matrix (GLGCM). By these spectral and textural features, probabilistic neural network (PNN) models for classification of fresh and frozen-thawed pork meats were established. Compared with the models using the optimum wavelengths only, optimum wavelengths with HS image features, and optimum wavelengths with GLCM image features, the model integrating optimum wavelengths with GLGCM gave the highest classification rate of 93.14% and 90.91% for calibration and validation sets, respectively. Results indicated that the classification accuracy can be improved by combining spectral features with textural features and the fusion of critical spectral and textural features had better potential than single spectral extraction in classifying fresh and frozen-thawed pork meat.

Non-destructive prediction of thiobarbituricacid reactive substances (TBARS) value for freshness evaluation of chicken meat using hyperspectral imaging.

This study examined the potential of hyperspectral imaging (HSI) for rapid prediction of 2-thiobarbituric acid reactive substances (TBARS) content in chicken meat during refrigerated storage. Using the spectral data and the reference values of TBARS, a partial least square regression (PLSR) model was established and yielded acceptable results with regression coefficients in prediction (Rp) of 0.944 and root mean squared errors estimated by prediction (RMSEP) of 0.081. To simplify the calibration model, ten optimal wavelengths were selected by successive projections algorithm (SPA). Then, a new SPA-PLSR model based on the selected wavelengths was built and showed good results with Rp of 0.801 and RMSEP of 0.157. Finally, an image algorithm was developed to achieve image visualization of TBARS values in some representative samples. The encouraging results of this study demonstrated that HSI is suitable for determination of TBARS values for freshness evaluation in chicken meat.

Combining the genetic algorithm and successive projection algorithm for the selection of feature wavelengths to evaluate exudative characteristics in frozen–thawed fish muscle

The potential use of feature wavelengths for predicting drip loss in grass carp fish, as affected by being frozen at -20°C for 24 h and thawed at 4°C for 1, 2, 4, and 6 days, was investigated. Hyperspectral images of frozen-thawed fish were obtained and their corresponding spectra were extracted. Least-squares support vector machine and multiple linear regression (MLR) models were established using five key wavelengths, selected by combining a genetic algorithm and successive projections algorithm, and this showed satisfactory performance in drip loss prediction. The MLR model with a determination coefficient of prediction (R(2)P) of 0.9258, and lower root mean square error estimated by a prediction (RMSEP) of 1.12%, was applied to transfer each pixel of the image and generate the distribution maps of exudation changes. The results confirmed that it is feasible to identify the feature wavelengths using variable selection methods and chemometric analysis for developing on-line multispectral imaging.

Development of hyperspectral imaging coupled with chemometric analysis to monitor K value for evaluation of chemical spoilage in fish fillets

K value is an important freshness index widely used for indication of nucleotide degradation and assessment of chemical spoilage. The feasibility of hyperspectral imaging (400-1000 nm) for determination of K value in grass carp and silver carp fillets was investigated. Partial least square (PLS) regression and least square support vector machines (LS-SVM) models established using full wavelengths showed excellent performances and the PLS model was better with higher determination coefficients of prediction (R(2)P = 0.936) and lower root mean square errors of prediction (RMSEP = 5.21%). The simplified PLS and LS-SVM models using the seven optimal wavelengths selected by successive projections algorithm (SPA) also presented good performances. The spatial distribution map of K value was generated by transferring the SPA-PLS model to each pixel of the images. The current study showed the suitability of using hyperspectral imaging to determine K value for evaluation of chemical spoilage and freshness of fish fillets.

Suitability of hyperspectral imaging for rapid evaluation of thiobarbituric acid (TBA) value in grass carp (Ctenopharyngodon idella) fillet.

The suitability of hyperspectral imaging technique (400-1000 nm) was investigated to determine the thiobarbituric acid (TBA) value for monitoring lipid oxidation in fish fillets during cold storage at 4°C for 0, 2, 5, and 8 days. The PLSR calibration model was established with full spectral region between the spectral data extracted from the hyperspectral images and the reference TBA values and showed good performance for predicting TBA value with determination coefficients (R(2)P) of 0.8325 and root-mean-square errors of prediction (RMSEP) of 0.1172 mg MDA/kg flesh. Two simplified PLSR and MLR models were built and compared using the selected ten most important wavelengths. The optimised MLR model yielded satisfactory results with R(2)P of 0.8395 and RMSEP of 0.1147 mg MDA/kg flesh, which was used to visualise the TBA values distribution in fish fillets. The whole results confirmed that using hyperspectral imaging technique as a rapid and non-destructive tool is suitable for the determination of TBA values for monitoring lipid oxidation and evaluation of fish freshness.

Spectral absorption index in hyperspectral image analysis for predicting moisture contents in pork longissimus dorsi muscles.

Spectral absorption index was proposed to extract the morphological features of the spectral curves in pork meat samples (longissimus dorsi) under the conditions including fresh, frozen-thawed, heated-dehydrated and brined-dehydrated. Savitzky-Golay (SG) smoothing and multiplicative scatter correction (MSC) were used for calibrating both the spectral reflectance and absorbance values. The absorption values were better than the reflectance values and the calibrated spectra by MSC were better than the raw and SG smoothing corrected spectra in building moisture content predictive models. The optimized partial least square regression (PLSR) model attained good results with the MSC calibrated spectral absorption values based on the spectral absorption index features (R(2)P=0.952, RMSEP=1.396) and the optimal wavelengths selected by regression coefficients (R(2)P=0.966, RMSEP=0.855), respectively. The models proved spectral absorption index was promising in spectral analysis to predict moisture content in pork samples using HSI techniques for the first time.

Combination of spectra and texture data of hyperspectral imaging for prediction of pH in salted meat

This study was carried out to investigate the feasibility of combining spectral with texture features in order to improve pH prediction for salted pork. Average spectra were extracted from the region of interest (ROI) of hyperspectral images over the wavelength region of 400-1000 nm and 9 characteristic spectral variables were then selected by principal components analysis (PCA). Meanwhile, gray-level gradient cooccurrence matrix (GLGCM) analysis was implemented on the first PC image (accounted for 96% of the total variance) to extract 13 textural feature variables. Partial least-squares regression (PLSR) was developed for predicting pH based on spectral, textural or combined data. Coefficient of determination (R(2)P) of 0.794 for the prediction samples based on data fusion was achieved, which was superior to the results based on spectra (R(2)P=0.783) or texture (R(2)P=0.593) alone. Hence, methods of combining spectral with texture analyses are effective for improving meat quality prediction.

Feasibility of using hyperspectral imaging to predict moisture content of porcine meat during salting process.

The feasibility of using hyperspectral imaging technique (1000-2500 nm) for predicting moisture content (MC) during the salting process of porcine meat was assessed. Different spectral profiles including reflectance spectra (RS), absorbance spectra (AS) and Kubelka-Munk spectra (KMS) were examined to investigate the influence of spectroscopic transformations on predicting moisture content of salted pork slice. The best full-wavelength partial least squares regression (PLSR) models were acquired based on reflectance spectra (Rc(2)=0.969, RMSEC=0.921%; Rc(2)=0.941, RMSEP=1.23%). On the basis of the optimal wavelengths identified using the regression coefficient, two calibration models of PLSR and multiple linear regression (MLR) were compared. The optimal RS-MLR model was considered to be the best for determining the moisture content of salted pork, with a Rc(2) of 0.917 and RMSEP of 1.48%. Visualisation of moisture distribution in each pixel of the hyperspectral image using the prediction model display moisture evolution and migration in pork slices.

Determination of trace thiophanate-methyl and its metabolite carbendazim with teratogenic risk in red bell pepper (Capsicumannuum L.) by surface-enhanced Raman imaging technique

Surface-enhanced Raman scattering (SERS) imaging coupling with multivariate analysis in spectral region of 200 to 1800cm-1 was developed to quantify and visualize thiophanate-methyl (TM) and its metabolite carbendazim residues in red bell pepper (Capsicum annuum L.). Least squares support vector machines (LS-SVM) and support vector machines (SVM) models based on seven optimized characteristic peaks that showed SERS effects of TM and its metabolite carbendazim residues were employed to establish prediction models. SERS spectra with first derivative (1st) and second derivative (2nd) method were subsequently compared and the optimized model of 1st-LS-SVM acquired showed the best performance (RPD=6.08, R2P=0.986 and RMSEP=0.473). The results demonstrated that SERS imaging with multivariate analysis had the potential for rapid determination and visualization of the trace TM and its metabolite carbendazim residues in complex food matrices.

Prediction of total volatile basic nitrogen contents using wavelet features from visible/near-infrared hyperspectral images of prawn (Metapenaeus ensis).

A visible/near-infrared hyperspectral imaging (HSI) system (400-1000 nm) coupled with wavelet analysis was used to determine the total volatile basic nitrogen (TVB-N) contents of prawns during cold storage. Spectral information was denoised by conducting wavelet analysis and uninformative variable elimination (UVE) algorithm, and then three wavelet features (energy, entropy and modulus maxima) were extracted. Quantitative models were established between the wavelet features and the reference TVB-N contents by using three regression algorithms. As a result, the LS-SVM model with modulus maxima features was considered as the best model for determining the TVB-N contents of prawns, with an excellent RP(2) of 0.9547, RMSEP=0.7213 mg N/100g and RPD=4.799. Finally, an image processing algorithm was developed for generating a TVB-N distribution map. This study demonstrated the possibility of applying the HSI imaging system in combination with wavelet analysis to the monitoring of TVB-N values in prawns.