Clustering Techniques for Marbles Classification

TL;DR

This paper explores various segmentation and clustering techniques to automate the classification of marbles based on visual appearance, addressing challenges posed by diverse colors and subjective human evaluation.

Contribution

It evaluates segmentation methods and applies clustering algorithms like LVQ and simulated annealing for automatic marble classification.

Findings

Segmentation techniques vary in effectiveness.

Clustering algorithms can automate marble classification.

Performance measured with standard quality metrics.

Abstract

Automatic marbles classification based on their visual appearance is an important industrial issue. However, there is no definitive solution to the problem mainly due to the presence of randomly distributed high number of different colours and its subjective evaluation by the human expert. In this paper we present a study of segmentation techniques, we evaluate they overall performance using a training set and standard quality measures and finally we apply different clustering techniques to automatically classify the marbles. KEYWORDS: Segmentation, Clustering, Quadtrees, Learning Vector Quantization (LVQ), Simulated Annealing (SA).