Generation of synthetic training data for an HMM-based handwriting recognition system

Abstract

A perturbation model for generating synthetic text lines from existing cursively handwritten lines of text produced by human writers is presented. Our purpose is to improve the performance of an HMM-based off-line cursive handwriting recognition system by providing it with additional synthetic training data. Two kinds of perturbations are applied, geometrical transformations and thinning/thickening operations. The proposed perturbation model is evaluated under different experimental conditions.