Generic Automatic Recognition System for Handwritten Pashto Documents

Development of an automatic recognition system of handwritten Pashto documents is addressed. The goal of the Phase II work is to develop a prototype of the automatic recognition system useable for screening handwritten Pashto documents such as personal letters. The proposed work builds on a successful feasibility demonstration of a handwritten Urdu word recognition system under a recent Army SBIR project. The Urdu SBIR project has demonstrated the feasibility of building a generic recognition framework for non-Arabic languages using Arabic-style script based on the Hidden Markov Model (HMM) approach. We have developed the novel integrated Contourlet- and Graph-based feature extraction algorithm that outperforms the state-of-the-art baseline approaches. Using a large Urdu handwriting database created under the Phase II, an Urdu document recognition system was constructed with recognition performance approaching the Arabic handwriting recognition system reported under the MADCAT program. The proposed Pashto recognition system development will be based on the same HMM-based framework. We shall first create a large Pashto handwriting database and use it to develop a complete recognition engine for handwritten Pashto documents. At the end of the Phase II, we shall deliver the database as well as the handwritten Pashto recognition system prototype to the MFLTS program office.