Off-line recognition of handwritten Arabic words using multiple Hidden Markov Models

Hidden Markov Models (HMM) have been used with some success in recognising printed Arabic words. In this paper, a complete scheme for unconstrained Arabic handwritten word recognition based on a Multiple discriminant Hidden Markov Models is presented and discussed. The overall engine of this combination of a global feature scheme with an HMM module, is a system able to classify Arabic-Handwritten words and has been tested on one hundred different writers. The system first attempts to remove some of the variation in the images that do not affect the identity of the handwritten word. Next, the system codes the skeleton and edge of the word such that feature information about the strokes in the skeleton is extracted. Then, a classification process based on a rule-based classifier is used as that a global recognition engine to classify words into eight groups. Finally, for each group, the HMM approach is used for trial classification. The output is a word in the lexicon. A detailed experiment has been carried out, and successful recognition results are reported.