Contribution on Character and word Recognition for Handwritten Arabic Text

Abstract

The recognition of handwritten Arabic words remains a vast subject of research. The work presented in this memory is devoted to the evaluation of a handwriting Arabic recognition system. For word segmentation, a promising word based segmentation method was used. It relay on calculating an optimal distance, which separate words. The Generic Fourier Descriptor transformation was chosen to extract the characteristic of objects. GFD has three input parameters radial frequency , angular frequency and an image containing one centred object. For multi classification, linear support vector machine was used. The data used in our system is extracted from KHATT database, which is freely available for the purpose of research.