Bulk Raisin Classification using Gray Level Co-occurrence Matrix

Raisin is one of the most important agricultural products. In this study, by using the machine vision approach, the quality of bulk raisin was evaluated in two different conditions. In the first case, six classes of good and bad raisins mixture, and in the latter case, 15 classes of good, bad and woody raisins have been studied. Classification results with Linear Discriminate Analysis (LDA) and Support Vector Machine (SVM) showed that the best classification accuracy of 6 classes was obtained by linear SVM method with an accuracy of 85.55%. The results for classifying 15 classes including good, bad and wood showed that the best result was obtained by linear SVM method but with a lower accuracy of 63.55%. The results showed that the GLCM method was able to detect the class of raisin bulk product and could replace the expert in raisin processing plants.