LAP Lambert Academic Publishing Classification of Marble Using Image Processing Techniques: Image Classification Using Machine Learning

This study attempted to develop a Marble quality classification model by comparing color, texture, and ensemble of color and texture. The image processing technique based on combined texture and color features was examined to achieve good results. Averages of 60 pictures taken for each marble (grade A, grade B, grade C). A grayscale coexistence matrix used for texture and color histogram for color extraction. Five textures and six color features were extracted from each Marble. To build the models for the prediction K-NN, ANN is examined. Experimental results, the ANN model shows good outcomes with combined texture and color features methods. The average accuracy of 83.3% and 93.7% is achieved for KNN and ANN respectively. Marble fractures and vines of the images have a strong impact on the performance of the classifier.