A Nonlinear Normalization Metold for Off-line Handwritten Chinese Characters
Liu Zong-fan · Journal of Xinjiang Normal University · 2008
How to solve the shape distortion of handwritten Chinese characters is the key problem in off-line handwritten Chinese character recognition.The aim of normalization is to remedy the distortion of handwritten Chinese characters and to reduce the difference among characters in the same class.A nonlinear normalization method is proposed in this paper,which adjusts,on the one hand,the distribution of strokes to equalize the distribution density of strokes,and on the other the thickness of strokes to make the width of strokes identical.Experiments show that the dispersion among the characters in the same class produced by the method in this paper is smaller than that produced by other methods.That is to say,the normalization method proposed in this paper is an effective one.