Design, Implementation, and Testing of Perturbation Method for Handwritten Numeral Recognition

This report presents a new approach to o -line handwritten numeral recognition. From the concept of perturbation due to writing habits and instruments, we propose a recognition method which is able to account for a variety of distortions due to eccentric handwriting. The new approach constitutes a shift from the usual pattern recognition paradigm where normalisation is the rst step prior to feature extraction and classi cation. Normalisation is replaced by a set of perturbation processes modelling writing habits and instruments. As a result, the subsequent operations { feature extraction and classi cation { are independently applied for each perturbation type yielding a set of results that are eventually combined. We tested our method on two worldwide standard databases of isolated numerals, namely, CEDAR and NIST, and obtained 99:09% and 99:54% correct recognition rates at norejection level, respectively. The latter result was obtained by testing on more than 170000 numerals. CR