Evaluating clustering methods for classification of marble slabs in an automated industrial marble inspection system

Marble quality classification is a very important issue since export of marble slabs has important economic impact on natural stone industry. Considering the massive transportation costs and requirements of marbles, it is necessary to achieve acceptably high performance in quality classification. However, classification of marble slabs in terms of quality is generally performed manually using human operators. This is rather subjective and prone to errors. Hence, automated and computerized methods are needed in order to obtain reproducible and objective results. In this work, we test the performance of clustering based classification on texture based feature sets. This is done in order to compare performances of different clustering techniques, which might be employed instead of neural networks, especially when there is not enough number of samples to construct a training set. Different clustering methods have been applied to our data set and the results are evaluated by using several cluster validation techniques.