Prediction of pork color attributes using computer vision system.

Color image processing and regression methods were utilized to evaluate color score of pork center cut loin samples. One hundred loin samples of subjective color scores 1 to 5 (NPB, 2011; n=20 for each color score) were selected to determine correlation values between Minolta colorimeter measurements and image processing features. Eighteen image color features were extracted from three different RGB (red, green, blue) model, HSI (hue, saturation, intensity) and L*a*b* color spaces.