摘要
本文提出了一种针对文本图象的平滑细化方法,该方法以汉字的结构特征为基础,由平滑和细化两大部分组成。平滑方法依据近邻象素相关函数,滤除了影响汉字结构特征的噪声。细化方法又分为快速边缘检测算法和内外边缘同步细化算法,该算法的显著特点,是利用汉字边缘特征,状态字和连通图指导快速细化。同时采用判别树形式使条件判断所需时间最少。该处理方法失真小,速度快。
This paper presents a new fast smoothing &. thinning method for text image.TheThis paper presents a new fast smoothing &. thinning method for text image.The method is based on the structural features of characters and contains two parts. the smoothing method and the, thinning method. The smoothing method based on relative discriminant function in the near neighbourhood of pixels largely removes the noise which has a harmful effect on character structure. The thinning method consists of two parts. the fast edge detection method and t-he synchronous thinning method of inside and outside edges of characters. A remarkable feature of this method is the use of character edges, status words and connected graphs to direct the fast thinning processing. Two decision tree classifiers are used to speed conditional decision procedure.This method has a high processing speed and fidelity.
出处
《中文信息学报》
CSCD
1990年第2期49-54,共6页
Journal of Chinese Information Processing
