GRADING OF MUSHROOMS USING A MACHINE VISION SYSTEM

The qualify features of the common white Agaricus bisporus mushroom we re quantified using image analysis in order to inspect and grade the m ushrooms by an automated system. The features considered were color, s hape, stem cut, and cap veil opening. Two human inspectors evaluated s amples which were divided into training and test sets. The vision syst em was trained to classify mushrooms into two quality grades using thr esholding. The human inspection results were compared with each other as well as the computer vision system results. Misclassification by th e vision system ranged from 8 to 56% depending upon the quality featur e evaluated, but averaged about 20%. The disagreement between inspecto rs ranged from 14 to 36%.