Training of On-Line Handwriting Text Recognizers with Synthetic Text Generated Using the Kinematic Theory of Rapid Human Movements

A method for automatic generation of synthetic handwritten words is presented which is based in the Kinematic Theory and its Sigma-lognormal model. To generate a new synthetic sample, first a real word is modelled using the Sigma-lognormal model. Then the Sigma-lognormal parameters are randomly perturbed within a range, introducing human-like variations in the sample. Finally, the velocity function is recalculated taking into account the new parameters. The synthetic words are then used as training data for a Hidden Markov Model based on-line handwritten recognizer. The experimental results confirm the great potential of the kinematic theory of rapid human movements applied to writer adaptation.

[1]  Réjean Plamondon,et al.  Synthetic on-line signature generation. Part I: Methodology and algorithms , 2012, Pattern Recognit..

[2]  Christian O'Reilly,et al.  Recent developments in the study of rapid human movements with the kinematic theory: Applications to handwriting and signature synthesis , 2014, Pattern Recognit. Lett..

[3]  Julian Fiérrez,et al.  Synthetic on-line signature generation. Part II: Experimental validation , 2012, Pattern Recognit..

[4]  Alejandro Héctor Toselli,et al.  On-Line Handwriting Recognition System for Tamil Handwritten Characters , 2007, IbPRIA.

[5]  R Plamondon,et al.  Studying the variability of handwriting patterns using the Kinematic Theory. , 2009, Human movement science.

[6]  M. Tahar Kechadi,et al.  Preprocessing Techniques for Online Handwriting Recognition , 2009, Intelligent Text Categorization and Clustering.

[7]  R. Plamondon,et al.  Impact of the principal stroke risk factors on human movements. , 2011, Human movement science.

[8]  Laurent Miclet,et al.  Synthetic On-line Handwriting Generation by Distortions and Analogy , 2007 .

[9]  Harris Drucker,et al.  Boosting Performance in Neural Networks , 1993, Int. J. Pattern Recognit. Artif. Intell..

[10]  Horst Bunke.  Template-based Synthetic Handwriting Generation for the Training of Recognition Systems , 2005 .

[11]  Frederick Jelinek,et al.  Statistical methods for speech recognition , 1997 .

[12]  Réjean Plamondon,et al.  A kinematic theory of rapid human movements , 1995, Biological Cybernetics.

[13]  Isabelle Guyon,et al.  UNIPEN project of on-line data exchange and recognizer benchmarks , 1994, Proceedings of the 12th IAPR International Conference on Pattern Recognition, Vol. 3 - Conference C: Signal Processing (Cat. No.94CH3440-5).

[14]  Réjean Plamondon,et al.  Learning handwriting with pen-based systems: computational issues , 2002, Pattern Recognit..

[15]  Horst Bunke,et al.  Generation and Use of Synthetic Training Data in Cursive Handwriting Recognition , 2003, IbPRIA.

[16]  Réjean Plamondon,et al.  The generation of handwriting with delta-lognormal synergies , 1998, Biological Cybernetics.

[17]  Hwan-Gue Cho,et al.  A New Synthesizing Method for Handwriting Korean Scripts , 1998, Int. J. Pattern Recognit. Artif. Intell..

[18]  Rafael Llobet,et al.  Training Set Expansion in Handwritten Character Recognition , 2002, SSPR/SPR.

[19]  Isabelle Guyon.  Handwriting Synthesis From Handwritten Glyphs , 1996 .

[20]  Claudio M. Privitera,et al.  A neural model for generating and learning a rapid movement sequence , 1996, Biological Cybernetics.

[21]  R. Plamondon,et al.  A multi-level representation paradigm for handwriting stroke generation. , 2006, Human movement science.

[22]  Réjean Plamondon,et al.  Development of a Sigma-Lognormal representation for on-line signatures , 2009, Pattern Recognit..