Perbandingan Klasifikasi Jenis Daging Menggunakan Ekstraksi Ciri Tekstur Gray Level Co-occurrence Matrices (GLCM) dan Local Binary Pattern (LBP)

Abstract

Meat has high nutritional value which is widely consumed by humans. The content contained in meat includes protein, vitamins, minerals, fats, and other substances that are needed by the body so that it can carry out activities every. However, unfortunately not all people can distinguish these types of meat, because the texture and color are almost similar. This is also often used by irresponsible meat sellers by mixing these types of meat or with other types of meat to get a big profit. Even though not everyone can consume certain types of meat for reasons of illness. This research was conducted to compare the GLCM and LBP methods for image classification of meat types based on texture analysis. The types of meat images that are classified are goat meat, buffalo meat, and horse meat. Image data is taken manually using a digital camera the Nikon D3200. The image was taken at a distance of 20 cm. Data testing and training was carried out using the Support Vector Machine (SVM) method. The results of the image classification accuracy of goat, buffalo, and horse meat using the GLCM method were 75.6%. While the results of the classification accuracy using the LBP method amounted to 85,6%. Thus, the LBP texture feature extraction method is recommended for classification of meat types using texture characteristics.