Visible hyperspectral imaging for predicting intra-muscular fat content from sheep carcasses

Intramuscular fat (IMF) content plays a key role in the quality attributes of meat, such as sensory properties and health considerations. The tenderness, flavour and juiciness of meat are examples of sensory attributes influenced by IMF content. Traditionally, IMF content in meat was determined using destructive, time consuming and at times unsuitable methods in industry applications. However, with recent advancement of technology, there has been an interest in exlporing ways to ascertain meat quality without damage. Hyperspectral imaging analysis is an emerging technology that combines the use of spectroscopy and computer imaging analysis to obtain both the spectral and spatial information of objects of interest. Hyperspectral imaging was initially developed for remote sensing, but has recently emerged as powerful tool for non-destructive analysis of quality in the food industry and has had very accurate results in the prediction of meat qualities such as IMF content. In this thesis, we use a data set of 101 hyperspectral images of sheep carcasses to investigate the ability of multivariate statistical methods to accurately predict IMF content