Share Email Print

Journal of Electronic Imaging

New baseline correction algorithm for text-line recognition with bidirectional recurrent neural networks
Author(s): Olivier Morillot; Laurence Likforman-Sulem; Emmanuèle Grosicki
Format Member Price Non-Member Price
PDF $20.00 $25.00

Paper Abstract

Many preprocessing techniques have been proposed for isolated word recognition. However, recently, recognition systems have dealt with text blocks and their compound text lines. In this paper, we propose a new preprocessing approach to efficiently correct baseline skew and fluctuations. Our approach is based on a sliding window within which the vertical position of the baseline is estimated. Segmentation of text lines into subparts is, thus, avoided. Experiments conducted on a large publicly available database (Rimes), with a BLSTM (bidirectional long short-term memory) recurrent neural network recognition system, show that our baseline correction approach highly improves performance.

Paper Details

Date Published: 21 June 2013
PDF: 12 pages
J. Electron. Imaging. 22(2) 023028 doi: 10.1117/1.JEI.22.2.023028
Published in: Journal of Electronic Imaging Volume 22, Issue 2
Show Author Affiliations
Olivier Morillot, Telecom ParisTech (France)
Laurence Likforman-Sulem, Telecom ParisTech (France)
Emmanuèle Grosicki, Délégation Générale pour l'Armement (France)

© SPIE. Terms of Use
Back to Top