SPECIAL ISSUE: Connecting Graphonomic Research and Pattern Recognition; Edited by H.-L. Teulings and A. W. A. Van GemmertNo Access

OFFLINE HANDWRITING RECOGNITION USING SYNTHETIC TRAINING DATA PRODUCED BY MEANS OF A GEOMETRICAL DISTORTION MODEL

T. VARGA

Institut für Informatik und angewandte Mathematik, Universität Bern, Neubrückstrasse 10, CH-3012 Bern, Switzerland

Search for more papers by this author

and

H. BUNKE

Institut für Informatik und angewandte Mathematik, Universität Bern, Neubrückstrasse 10, CH-3012 Bern, Switzerland

Search for more papers by this author

https://doi.org/10.1142/S0218001404003666Cited by:9 (Source: Crossref)

Abstract

A perturbation model for the generation of synthetic textlines from existing cursively handwritten lines of text produced by human writers is presented. The goal of synthetic textline generation is to improve the performance of an offline cursive handwriting recognition system by providing it with additional training data. It can be expected that by adding synthetic training data the variability of the training set improves, which leads to a higher recognition rate. On the other hand, synthetic training data may bias a recognizer towards unnatural handwriting styles, which could lead to a deterioration of the recognition rate. In this paper the proposed perturbation model is evaluated under several experimental conditions, and it is shown that significant improvement of the recognition performance is possible even when the original training set is large and the textlines are provided by a large number of different writers.

References

H. S. Baird, The state of the art of document image degradation modeling, in Proc. 4th IAPR Workshop on Document Analysis Systems (DAS 2000), Dcember 2000, Rio de Janeiro, Brazil . Google Scholar
J. Cano, J. C. Pérez-Cortes, J. Arlandis and R. Llobet, Training set expansion in handwritten character recognition, in Proc. 9th SSPR/4th SPR 2002, Windsor, Ontario, Canada, pp. 548–556 . Google Scholar
H. Drucker, R. Schapire and P. Simard, Advances in Neural Information Processing Systems 5, eds. S. Hansonet al. (Morgan Kaufmann, San Mateo, CA, 1993) pp. 42–49. Google Scholar
I. Guyon, Handwriting synthesis from handwritten glyphs, in Proc. 5th Int. Workshop Frontiers in Handwriting Recognition 1996, Essex, England, pp. 309–312 . Google Scholar
S. Günter and H. Bunke, Optimizing the number of states, training iterations and Gaussians in an HMM-based handwritten word recognizer, in Proc. 7th Int. Conf. Document Analysis and Recognition August 2003, Edinburgh, Scotland, pp. 472–476 . Google Scholar
T. M. Ha and H. Bunke, IEEE Trans. Patt. Anal. Mach. Intell. 19(5), 535 (1997). Crossref, Web of Science, Google Scholar
A. Manzanera and T. M. Bernard, Improved low complexity fully parallel thinning algorithm, in Proc. 10th Int. Conf. Image Analysis and Processing 1999, Venice, Italy, pp. 215–220 . Google Scholar
J. Mao and K. M. Mohiuddin, Patt. Recogn. Lett. 18, 1415 (1997). Crossref, Web of Science, Google Scholar
U.-V. Marti and H. Bunke, Int. J. Pattern Recognition and Artificial Intelligence 15(1), 65 (2001). Link, Web of Science, Google Scholar
U.-V. Marti and H. Bunke, Int. J. Docum. Anal. Recogn. 5(1), 39 (2002). Crossref, Google Scholar
M. Mori, A. Suzuki, A. Shio and S. Ohtsuka, Generating new samples from handwritten numerals based on point correspondence, in Proc. 7th Int. Workshop on Frontiers in Handwriting Recognition 2000, Amsterdam, The Netherlands, pp. 281–290 . Google Scholar
L. Rabiner and B.-H. Juang , Fundamentals of Speech Recognition ( Prentice Hall , 1993 ) . Google Scholar
H. A. Rowley, M. Goyal and J. Bennett, The effect of large training set sizes on online Japanese Kanji and English cursive recognizers, in Proc. 8th Int. Workshop on Frontiers in Handwriting Recognition 2002, Niagara-on-the-Lake, Ontario, Canada, pp. 36–40 . Google Scholar
S. Setlur and V. Govindaraju, Patt. Recogn. Lett. 15, 901 (1994). Crossref, Web of Science, Google Scholar
P. Simard, Y. Cun and J. Denker, Advances in Neural Information Processing Systems 5, eds. S. Hansonet al. (Morgan Kaufmann, San Mateo CA, 1993) pp. 50–58. Google Scholar
P. Soille , Morphological Image Analysis ( Springer-Verlag , Berlin Heidelberg , 1999 ) . Crossref, Google Scholar
T. Varga and H. Bunke, Generation of synthetic training data for an HMM-based handwriting recognition system, in Proc. 7th Int. Conf. Document Analysis and Recognition August 2003, Edinburgh, Scotland, pp. 618–622 . Google Scholar
T. Varga and H. Bunke, Effects of training set expansion in handwriting recognition using synthetic data, Proc. 11th Conf. Int. Graphonomics Soc., November 2003, Scottsdale, Arizona, USA, pp. 200–203 . Google Scholar
T. Varga and H. Bunke, Offline handwritten textline recognition using a mixture of natural and synthetic training data, in Proc. 17th Int. Conf. Pattern Recognition 23–29 August 2004, Cambridge, UK, Vol. 11, pp. 545–549 . Google Scholar