ŞİŞECİ, Melike Süleyman Demirel Üniversitesi - Bilgisayar Mühendisliği Bölümü, Turkey , CETİŞLİ, Bayram Süleyman Demirel Üniversitesi - Bilgisayar Mühendisliği Bölümü, Turkey

Determination of Travertine Plate Stone Classes with K-means and Fuzzy Cmeans Clustering Methods

25290

It is aesthetically important that surfaces appear as homogeneous in the surface coverings made of natural stones. At this point, there are classes determined by both the customers’ demands and the manufacturers of stone. During the production, since the classification of stone plates (selection) is generally done manually, it is not possible to reach a standard. In addition, the closest class is determined and plates are divided into these classes in case stones are non-compliance with pre-determined class standards. At this point, an objective classification should be done being independent from people’s different visuality. In this study, K-means which is one of the learning models without a teacher and fuzzy C-means methods are compared with each other, new class definition is foreseen for travertine slabs. In the methods, analyzing both color and homogeneity, the most suitable qualities have been determined for travertine’s. Experimental studies show that the discrimination between the proposed new classification forms and travertine classes is more obvious and the distribution of them in classes seems more pleasant.

238

K , means clustering methods , Fuzzy C , means , Homogeneity analysis , Color analysis , Travertine classification , Image Processing.

Journal Of Natural an‎d Applied Sciences

247