An Improved Method of Detecting Pork Freshness Based on Computer Vision in On-line System

On the basis of computer vision, this paper studied and developed an on-line detection system for pork freshness, which include the overall design of the system scheme, the hardware design and functions, the software functions and detection algorithm. The systematical hardware is composed of image acquisition unit, light source unit, control unit, drive transmission device and computer. For the software implementation of the collected images, the processes include three steps as following: 1) Otsu algorithm was applied to remove the disturbance of background and other noises. 2) The fat areas were eliminated according to color difference between pork fat and the muscle. 3) A new Color Region Ratio (CRR) feature extraction method applying colorlayering approach was proposed for the identification of pork freshness. The testing of 100 samples have shown that the CRR feature is highly correlated with pork freshness, reaching 88 % detection accuracy, and it is feasible to use CRR feature to detect the pork freshness. Copyright © 2014 IFSA Publishing, S. L.