KNN Algorithm Optimization in GLCM-Based Beef and Pork Image Classification

Beef is the most consumed meat by the people of Indonesia. The increase in beef prices that are getting higher from year to year makes beef producers commit fraud, namely by mixing beef with beef. This makes consumers feel aggrieved and get meat that is not as expected. So a system is needed to solve the problem, which is able to distinguish the type of beef and the type of pork. Therefore, this study proposes a method used to identify pork and beef based on imagery. There are several stages to identify pork and beef, namely by segmenting or separating objects with the background. The Otsu method is used for the segmentation process. Objects that have been segmented are carried out the extraction process of RGB and HSV color characteristics. Then proceed with the feature extraction process with GLCM. The classification process is carried out with the KNN algorithm with the best K value search optimization using the Elbow method. The results showed that this study was able to identify pork and beef with a testing accuracy of 91.66%. While the most optimal K value is obtained using the Elbow method of K = 7 with an error value = 0.08. The highest score for recall at K=7 was 88.79%, the highest value for precision was 97.17% at K=7, and the highest score for F1 Score was 92.79% at K=7.