摘要
提出了一个跟踪手写体字符骨架的新方法，该方法能够从静态字符骨架中提取出由笔落到笔起的完整笔划，也就是说，能够在离线情况下重新获得部分已丧失的书写字符时的动态信息．从而使得字符的离线识别能够采用在线识别的特征（或部分特征），提高离线识别率．这里提出的笔划延续最大成势方向和定向搜索邻域的概念解决了笔划在交叉点处的跟踪问题．本方法简单、直观，符合人的视觉．实验结果表明了该方法的有效性．
A new method tracing the skeleton of handwriting characters is proposed. This method makes it possible to obtain the complete strokes between the pen down and pen up from the skleleton of static character(that is to say to recover part of the lost script dynamics), thus the features or part of the features available in the on line case can be exploited for the off line recognition to increase the recognition rate. Here the concepts of most likely curve extending direction and oriented search neighborhood suggested can solve the problem of tracing strokes at joints. This method is simple and audiovisual and conforms to a persons vision. The experimental results indicate the efficiency of our method.
出处
《兰州大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
1998年第1期41-46,共6页
Journal of Lanzhou University(Natural Sciences)
关键词
细化
离线识别
笔划跟踪
完整笔划提取
字符骨架
OCR thinning dynamic information off line recognition distortion stroke tracing complete stroke
