Explore UCD

UCD Home >

Research Projects in 2015-2016

Wednesday, 28 November, 2018

Hyperspectral imaging technology for the quality inspection of fish products (SPECTRAFISH) (Postdoc)

Riccioli C and Da-Wen Sun

Sponsors: European Commission, 7th Framework Programme Theme Capacities

In recent years, hyperspectral imaging (HSI) has gained a wide recognition as a nondestructive and fast quality and safety analysis and assessment method for a wide range of food products. The SPECTRAFISH project aims to bridge the lack of rapid and objective methods of noninvasively inspecting based on quality and safety attributes of finfish products. The system will integrate two conventional optical sensing technologies of computer vision and near infrared spectroscopy into unique imaging sensors, a hyperspectral imaging system that can provide not only spatial but also spectral information for each pixel in an image. The development of a hyperspectral imaging device for the automatic, rapid, objective and non-invasive measurement of quality and safety attributes of finfish fillets will be carried out throughout the project.


Application of hyperspectral imaging technique for measurement of external defects of potatoes (PhD)

Su WH, He HJ and Da-Wen Sun

Sponsors: CSC-UCD Scholarship Scheme

Hyperspectral imaging (HSI) techniques in the wavelength of 400−1000 nm were applied for the rapid and non-destructive measurement of external defects of potatoes. The hyperspectral images of seven potato types were obtained. Then the reflectance spectrums of the interested areas of potato in these hyperspectral images were extracted and analysed. Five feature wavelengths (478, 670, 723, 819 and 973 nm) were selected based on principal component analysis. Principal component analysis was conducted again based on the five selected characteristic wavelengths. Potato external defects were identified through image processing methods, such as threshold segmentation, corrosion, expansion and connectivity analysis. The correct recognition rate of all the seven potato types using principal component analysis method of the characteristic wavelengths achieved 82.50%. The results showed that hyperspectral imaging technique was suitable for rapid and non-destructive assessment of external defects of potatoes.


Application of near-infrared hyperspectral imaging for non-destructive determination of 2-thiobarturic acid (TBA) value in Atlantic salmon (salmo salar) fillets (PhD)

Xu JL, Riccioli C and Da-Wen Sun

Sponsors: CSC-UCD Scholarship Scheme

This study investigated the potential of using hyperspectral imaging technique in the near infrared region (900–1700 nm) for rapid and non-invasive determination of 2-thiobarbituric acid (TBA) value for monitoring lipid oxidation in Atlantic salmon (Salmo salar) fillets during cold storage. Hyperspectral cubes were acquired at different storage stages and their corresponding spectra data were extracted. Partial least square regression (PLSR) calibration models were established with full spectral region between the spectral data and the reference TBA values. Good performance for predicting TBA value was observed with determination coefficients (rp) of 0.85 and root-mean-square errors of prediction (RMSEP) of 2.24 μmol MDA/kg fish. The research demonstrated that hyperspectral imaging technique is suitable for rapid and non-destructive evaluation of lipid oxidation in salmon flesh during cold storage.


Moisture content distribution in mango slices during microwave-vacuum drying using NIR hyperspectral imaging (PhD)

Pu YY and Da-Wen Sun

Sponsors: CSC-UCD Scholarship Scheme

Dried mango slices were obtained by using a microwave-vacuum drying system. Moisture content of the mango slices during the drying process was analyzed. Spectral and spatial information of each mango slice was acquired by a lab-scale NIR hyperspectral imaging system. The Page model, with a higher fitting precision of R2=0.978, was found to describe the current drying process. For moisture content prediction, partial least square (PLS) was applied to correlate the mean spectrum and the referenced moisture content of each mango slice. Feature wavebands were extracted for moisture content prediction by stepwise regression (SR) and competitive adaptive reweighted sampling (CARS). Prediction performance of PLS models based on full-wavelength range (FW) and the selected important wavebands were compared. The coefficients of determination (R2) and the root mean square errors for prediction (RMSEP) of FW-PLS model, SR-PLS model and CARS-PLS model were 0.969, 0.962, 0.971 and 4.796%, 5.340%, 4.891%, respectively. The simplified model CARS-PLS was implemented into the moisture visualization procedure. The present study demonstrated that hyperspectral imaging was a useful tool for non-destructively and rapidly measuring and visualizing the moisture content during drying process.


Visualization of pseudomonas loads of salmon flesh using near-infrared hyperspectral imaging technique (PhD)

He HJ and Da-Wen Sun

Sponsors: CSC-UCD Scholarship Scheme

The feasibility of near-infrared hyperspectral imaging in the range of 900−1700 nm (NIR) for predicting Pseudomonas loads on salmon flesh were investigated. Hyperspectral images of salmon samples stored at different days were acquired and spectral information was extracted to relate to reference values of Pseudomonas loads measured by traditional Pseudomonas CFC-selective medium, using partial least square (PLS) regression algorithm. The quantitative relationship between the spectra and the measured Pseudomonas loads was established, leading to a PLS model with correlation coefficient of prediction (RP) of 0.95 and root mean square error of prediction (RMSEP) of 0.52. Competitive adaptive reweighted sampling (CARS) algorithm was applied to select the most important wavelengths (941, 1105, 1161, 1178, 1222, 1242, 1359, 1366, 1628 and 1652 nm) to simplify the PLS model. With the ten important wavelengths, a CAR-PLS model was developed with RP of 0.95 and RMSEP of 0.49. By transferring the CARS-PLS model to each pixel of hyperspectral images of samples, Pseudomonas load distribution was mapped to display the spatial variation of Pseudomonas load from sample to sample and spot to spot within the same sample. The overall results indicated that NIR hyperspectral imaging technique has a great potential and could be used for determining Pseudomonas loads on salmon flesh during the cold storage.


Comparison of three desorption isotherm determination methods on by using microcrystalline cellulose (PhD)

Zhang L, Zhang Z, Drummond L and Da-Wen Sun

Sponsors: European Commission - 7th Framework Programme, CSC-UCD Scholarship Scheme


Application of hyperspectral imaging on quality assessment of fruits (PhD)

Wen L, Pu YY and Da-Wen Sun

Sponsors: CSC-UCD Scholarship Scheme

Hyperspectral imaging (HSI) has gained wide recognition as a rapid, chemical-free, and non-destructive quality and safety analysis method for a wide range of food products, by simultaneously offering both spatial information and spectral signals from one object. This study focuses on the recent applications of hyperspectral imaging on quality assessment for fruits. First, the fundamental principles, major instrumental components and data analysis methods of hyperspectral imaging are presented. Second, applications of hyperspectral imaging on quality inspection of fruits are reviewed specifically summarized into bruise, moisture content and fruit fly/insect infestation.


Application of hyperspectral imaging for automatic differentiation of organically and conventionally farmed salmon (PhD)

Xu JL and Da-Wen Sun

Sponsors: CSC-UCD Scholarship Scheme

This study was carried out to investigate the potential of short-wave near infrared (400-1000 nm) hyperspectral imaging (HSI) system for differentiation of organic and conventional farm-raised salmon fillets in fresh and chill-stored conditions. Hyperspectral cubes were acquired and their corresponding spectra data were extracted. Principal component analysis (PCA) was applied to explore the variance between different classes of salmon. Partial least squares-discriminant analysis (PLS-DA) was used to build classification models for recognition and authentication of the tested samples. The best results were obtained with the correct classification rate (CCR) of 98.2% in validation and CCR of 97.1% in cross-validation, which suggested the capability of the hyperspectral imaging for objective and rapid categorization of the two salmon varieties under circumstances of fresh and chill-stored conditions.


Potential of time-series hyperspectral imaging for freshness detection in salmon during cold storage (PhD)

Zhao YM, Xu JL, and Da-Wen Sun

Sponsors: CSC-UCD Scholarship Scheme

Freshness is a primary attribute of consumers’ acceptance to evaluate quality of fish fillets. This experiment investigated the potential of using near infrared hyperspectral imaging (900-1700 nm) for rapid and non-destructive prediction of freshness in Atlantic salmon (salmon salar) fillets during cold storage. One salmon fillet was stored in the refrigerator at controlled temperature of -1℃ over 12 days. Hyperspectral images were obtained at 0, 3, 6, 9 and 12 days and their corresponding spectral data were extracted. Savitsky Golay smoothing and derivatives (SG) and standard normal variate (SNV) were used as spectral pretrement methods, principal components analysis (PCA) was applied for exploration of image variance. The results demonstrated that hyperspectral imaging technique has the potential to present and classify the differences of salmon samples with different storage days.


Effects of different drying methods on quality changes of fruits (MEngSc)

Zhang K, Pu YY and Da-Wen Sun

Sponsors: University College Dublin

Drying is a conventional food preservation method that inhibits the growth of microorganism in perishable products, such as fruits. The dried fruits qualities are influenced by different dehydration processes, and the most common quality damages are color change, nutrients loss and texture shrinkage. Generally, the hot air drying results in bad product quality, and the products dried by microwave drying are better than the hot air dried products but worse than the freeze dried products. Novel drying methods can lead to similar final quality with a lower process cost.


UCD Food Refrigeration & Computerised Food Technology

Food Refrigeration and Computerized Food Technology University College Dublin Agriculture & Food Science Centre Belfield, Dublin 4, Ireland.
T: +353 1 716 7342