New distance measures applied to marble classification

Abstract

Automatic marble classification based on their visual appearance is an important industrial issue, but due to the presence of randomly distributed high number of different colors and its subjective evaluation by human experts, the problem remains unsolved. In this paper, several new measures based on similarity tables built by human experts are introduced. They are used to improve the behavior of some clustering algorithms and to quantitatively characterize the results, increasing the correspondence of the measures to the visual appearance of the results. The obtained results show the effectiveness of the proposed methods. © Springer-Verlag Berlin Heidelberg 2003.