A fractal approach to predict fat content in meat images

Intramuscular content in meat influences some important meat quality characteristics. Chemical analysis is currently used to determine intramuscular fat percentage in beef meat. Nevertheless, this is a tedious and expensive technique. For the food industry, it will be very useful to have a cheaper and non-destructive technique to determine fat content. We investigate the feasibility of a new method to predict fat content. We model meat structure as a fractal, and assume the projected image can be described by a fractional Brownian motion (FBM). Experimental results show that this assumption is satisfied over an acceptable scale range. The Hurst coefficient of the FBM appears to present a high correlation with fat percentage.