Determination of doneness of beef steaks using fuzzy pattern recognition

The c-means and k-nearest neighbor algorithms of pattern recognition were used as methods of determining the degree of doneness of beef ribeye steaks. For each method, an approach employing the theory of fuzzy sets was used in addition to the traditional crisp approach. Steaks were heat processed to various degrees of doneness. The red, green and blue components of the steak images were acquired using a digital image processing system. Values were then converted to the hue, intensity and saturation. The use of the mean and standard deviation of these six values as features was investigated. The means of the red, intensity and saturation and the standard deviation of the red were found to be of most value as features for the classification techniques investigated.