Use of hyperspectral imaging technologies for prediction of beef meat quality

As an emerging technology, hyperspectral imaging (HSI) providesa unique non-destructive way of analysing food quality. In the current application HSI is applied to meat quality analysis, based on an image cube captured at different wavelengths, which usually covers from visible (VIS) to near infrared (NIR) bands. Many researchers have found that there is a relationship between eating quality of beef and corresponding sensory properties such as tenderness and flavour. The tenderness can be assessed by measuring the slice shear force (SSF) and the ultimate pH value is an important shelf-life and colour parameter. In this project, HSI has been employed to predict the SSF measurement and pH value of captured beef samples at 7 days and 14 days post mortem and the results are compared with the existing NIR reflectance spectroscopy. Principal component analysis (PCA) is employed for feature extraction and selection with support vector machine (SVM) used for the prediction.>600 beef M. longissimusthoracissamples at 48 hours post mortemhave been scanned in three abattoirs (200 per abattoir over two consecutive days), using both hyperspectral imaging system and NIR reflectance spectroscopy. SSF and pH measures of steaks were collected by QMS. Preliminary results show that both HSI and NIR predict pH value more successfully than SSF. For SSF prediction, HSI (visible bands only)shows great potential as it yields higher coefficient of determination R2 than NIR. For the pH value prediction, the coefficient of determination (R2 ) of HSI is also higher than that of NIR. This indicates that HSI techniques can be more favourablethan NIR reflectance spectroscopy for accurate prediction of beef SSF and ultimate pH.