摘要
为了解决字符识别过程中笔划不易跟踪及其若干特征难以提取等问题,给出了一种新的适合于笔划跟踪的并行细化算法。该算法运用井行机理,在满足细化基本要求的前提下,去掉了不便于笔划跟踪的冗余点,细化速度快且质量较好。该算法已成功地运用于手写数字识别系统及汉字识别系统中,其中利用该算法的表格识别系统已经达到实用水平。
in some OCR systems, there exist a few problems. Such as strokes can not be traced and some features can't be extracted easily. In order to solve these problems, a parallel thinning algorithm is develoPed and proposed. With this algorithm,each image can be skeletonned fast and perfectly. Furthermore, strokes of the images dealed with this method can be easily traced to extract some important features. This algorithm has been used in several OCR systems successfully.
出处
《哈尔滨工业大学学报》
EI
CAS
CSCD
北大核心
1994年第4期42-45,共4页
Journal of Harbin Institute of Technology
关键词
模式识别
笔划跟踪
并行算法
Pattern recognition
skeletonization
stroke tracing