Integration GLCM and geometric feature extraction of region of interest for classifying tuna

Image of tuna as bigeye, skipjack and yellowfin have very high color similarity, but in the texture and shape can be differentiated. It requires a method to perform feature extraction of bigeye, skipjack and yellowfin appropriately, so the results on a classification of tuna have a high accurate rate. We propose a method to integrate gray level co-occurrence matrix (GLCM) and geometric feature extraction of region of interest (ROI) for classifying tuna. To measure the texture of tuna is require making region in an image using centroid as a parameter of center boundary to help determine head, body and tail. Thus, maximally get its extraction and produce an accurate classification. The experiment results show the integration GLCM and geometric shape feature extraction is successful and classify very well the image of bigeye, skipjack and yellowfin with 86.67% accurate, 0.8% Kappa, 0.11% MAE, 0.28% RMSE, 24.71% RAE and 58.95% RRSE using 10-fold cross-validation of the entire dataset.