Construction of language models for an handwritten mail reading system

Olivier Morillot (1) , Laurence Likforman-Sulem (1) , Emmanuèle Grosicki (2)

1 LTCI - Laboratoire Traitement et Communication de l'Information
2 DGA/CTA/DT/GIP - CEP Arcueil

This paper presents a system for the recognition of unconstrained handwritten mails. The main part of this system is an HMM recognizer which uses trigraphs to model contextual information. This recognition system does not require any segmentation into words or characters and directly works at line level. To take into account linguistic information and enhance performance, a language model is introduced. This language model is based on bigrams and built from training document transcriptions only. Different experiments with various vocabulary sizes and language models have been conducted. Word Error Rate and Perplexity values are compared to show the interest of specific language models, fit to handwritten mail recognition task.

Mots clés: Offline Handwriting recognition; handwritten mail; language modeling; Hidden Markov Models; text-line recognition; n-gram

Domaines: Traitement du texte et du document

Olivier Morillot, Laurence Likforman-Sulem, Emmanuèle Grosicki. Construction of language models for an handwritten mail reading system. IS&T/SPIE 24th Annual Symposium on Electronic Imaging - Document Recognition and Retrieval XIX, Jan 2012, San Francisco, United States. ⟨10.1117/12.911965⟩. ⟨hal-00737407⟩