Automated Classification of Marble Types Using Texture Features and Neural Networks: A Robust Approach for Enhanced Accuracy and Reproducibility

The automatic classification of marble is a challenging task that requires a combination of image processing, texture analysis, and machine learning techniques. However, the classification by human experts is error-prone and subjective. Therefore, automated computer-assisted procedures are required to obtain reproducible and objective results. In this paper, a new approach for classifying marble types is proposed based on their texture features extracted from images using various image processing techniques. Neural network models were used to train and classify the different types of marble based on their features. The proposed approach was tested on a dataset of 5088 images of different marble types and achieved high accuracy in classifying the marble types. The proposed method surpasses present-day automated and human methods in terms of accuracy, ranging from 97.3% to 100%, for a large number of types of marble. The results provide evidence that texture properties derived from the grayscale co-occurrence matrix can effectively distinguish between different varieties of marble. It is proven that textural cues and neural networks are efficient in the automated identification of marble species; this research data has practical applications in many industries that use marble.