摘要
手写体汉字变形问题是无约束手写体汉字识别的关键问题。规范化处理的目的是对字符图像点阵进行变换,矫正手写体汉字变形,减小同类字符之间的差异。文章提出了一种非线性规范化方法,它一方面调整了汉字的笔划分布,使笔划的分布密度趋于均匀;另一方面又调整了笔划的粗细,使笔划的宽度趋于一致。实验显示,用文章提出的方法产生的同类字符点阵的离散度小于用已有的几种非线性规范化方法产生的同类字符点阵的离散度,表明提出的非线性规范化方法,较之已有的几种非线性规范化方法更有效地减小了同类字符之间的差异。
How to solve the shape distortion of off-line handwritten Chinese character recognition. of handwritten Chinese characters and to reduce the handwritten Chinese characters is the key problem in The aim of normalization is to remedy the distortion difference among characters in the same class. A non-linear 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.
出处
《新疆师范大学学报（自然科学版）》
2008年第3期41-45,共5页
Journal of Xinjiang Normal University(Natural Sciences Edition)
关键词
脱机手写汉字识别
非线性规范化
离散度
Off-line handwritten Chinese character recognition
nonlinear normalization
dispersion
