Online Handprinted "Hiragana" Recognition using Vector Correlation

Abstract

Although a lot of research has been carried out on cursive character recognition, validity of matching measures has not been considered adequately. In online handprinted character recognition, the patterns are often represented by directional data or angular data. But, treatment of such data requires different operations from those of data on ordinary Euclidean space. In this paper, we propose a new matching measure to discriminate between a pair of directional data series, which represent respectively online handprinted character. We confirmed through the classification experiment that, the proposed measures are highly effective for online handprinted character recognition.