Browse by topic
Type of publication
Aligning handwritten text images and transcriptions of historic documents. In proceedings of the 2nd EVA 2008 Vienna Conference, 2008. pp. 87-94.A method to align handwritten text images and their respective (electronic) transcriptions based on the Viterbi algorithm is presented in this work. The approach takes advantage of the underlying segmentation made by Viterbi decoding in handwritten text recognition based on Hidden Markov Models. Two levels of alignments are carried out: word level and line level; the latter for pages transcribed without line break synchronization. According to various metrics used to measure the quality of the alignments, encouraging results are obtained.