Constraints on handwriting Korean characters to improve the machine readability

Realizing that the availability of a practical recognition system for Korean handwritten characters without any constraints has a long way to go, we have attempted to find a set of writing constraints which significantly improves the machine-readability. Based on our observation that the majority of the misrecognitions reported are caused by ambiguous characters, we have developed a set of writing constraints which maximally disambiguate those characters. Through experiments, we have confirmed that the recognition rate of those handwritten data could be improved significantly by applying the proposed set of constraints.