Automatic extraction of lean-tissue contours for beef quality grading

In evaluating beef quality, proper extraction of the boundary of lean-tissue from the section of beef rib is the crucial first step. The quality features of a beef carcass are determined by size, marbling state, and the colour of the lean-tissue on the 13th rib. Because of the inhomogeneous distribution and fuzzy pattern of the fat and lean-tissues on various beef-cuts, it is difficult to automatically extract a proper contour of the lean-tissue. In this research, image-processing algorithms have been developed to automatically extract the boundary of the lean-tissue in a robust way as human expert does. Algorithms include automatic threshold determination using Rényi entropy for beef-cut segmentation and contour modification, along with binary morphological approaches. The algorithms were applied to 36 beef-cut samples and successfully demonstrated contour extraction with average percentage error of 2.63% and average pixel distance error of 2.51 pixels when compared to results from a human expert. Computing time for the contour extraction process was around 5 s. A mobile image acquisition and processing system was also developed for on-site application. The system was composed of a hand-held image acquisition unit and a mobile computing unit mounted with a touch-pad screen. The algorithms developed were applied to the mobile processing unit as a supplementary grading device.