Cuneiform Symbols Recognition Based on K-Means and Neural Network

Abstract

Cuneiform is the ancient writing systemin the world. But, there is no clear interest recognition cuneiform symbol, despite its importance. This  research interested in building an algorithm for  cuneiform symbol recognition. Firstly, the Sumerian texts  were entered  through the scanner and make some initial preprocessing operations such as segmentation for the purpose of cutting the text and getting a cuneiform symbol. Then, features were extracted for each symbol by using vertical and horizontal projections, centre of gravity, and connected component. Because of the large number of cuneiform symbols, the similar symbols are clustered by K-means algorithm, then multilayer  neural networks are used to classify the symbol within the same cluster. The proposed algorithm  gave good results.