A stochastic representation of cursive Chinese characters for on-line recognition

In this paper, a stochastic representation of on-line Chinese characters of cursive style is proposed. A character in this representation is modeled by a sequence of concatenated stochastic curves, termed stochastic cubic Bezier curves (SCBC), with random noises. Furthermore, we also propose a curve alignment procedure to consistently match an input character with a stochastic reference one. Some classification experiments were performed. The stochastic approach is hardly sensitive to the characters with linked and degraded strokes; meanwhile, its recognition rate is higher than 95% even for deformed confusing characters. (C) 1997 Pattern Recognition Society.