Classification of ancient inscription images on the basis of material of the inscriptions

Machine Learning and AI has allowed us to process images and help us solve vital problems. In this paper, we are classifying inscription images into three image inscription classes namely stone inscriptions, metal inscriptions and palm leaves inscriptions. Due to decaying materials of ancient inscriptions, the classification of such materials becomes challenging. To address this problem, we are using various feature extraction methods like GLCM, KAZE, BRISK to implement texture based feature detection and subsequently classifying them. Both linear and non-linear methods for feature extraction are being used. The paper is concluded by performing classification of images and also making comparisons between all the feature extraction methods in terms of memory required by the algorithm, time scalability and accuracy score of the method. The paper consists of rich information which would be useful in decision making in classification problems.