Cursive on-line Handwriting word recognition using a bi-character model for large lexicon applications
Résumé
This paper deals with on-line handwriting recognition in a closed-world environment with a large lexicon. Several applications using handwriting recognition have been developed, but most of them consider a lexicon of limited size. Many difficulties, in particular confusions during the segmentation stage, are linked to the use of a large lexicon, with large writing variations and an increased complexity of the connections between characters. In order to circumvent these problems, we introduce in this paper an original method based on a new analytical approach using two levels of recognition models: an isolated character recognizer and an original bi-character recognition model. The idea behind the bi-character model is to recognize jointly two neighboring characters. The objective is to reduce the confusions between characters occurring during the segmentation step. Experiments show an interesting improvement of the recognition rate when introducing the bi-character model, as the recognition rate is increased of 7.2% for a 1000 words lexicon, of 9.1% for a 2000 words lexicon, and up to 15% for a 10000 words lexicon.