HMM-Based System for Transcribing Chinese Handwriting

A novel recognition strategy is proposed for the transcription of Chinese handwritten documents. The recognizer adapts continuous density Hidden Markov Model (HMM) as the recognition engine. It incorporates character segmentation and recognition in one step avoiding character segmentation phase. Textline is extracted and converted to observation sequence by sliding windows first. Then Baum-Welch algorithm is used to train character HMMs. Finally, best character string in maximizing a posteriori criterion is found out through Viterbi algorithm as output. Experiments are conducted on a writer-dependent Chinese handwriting database with a 1,695 lexicon. The results show that our baseline recognizer outperforms much one popular commercial handwritten character recognition product and the strategy presented in this paper is a promising research direction.