摘要: 以印刷体现代藏文白体、黑体、圆体、长体、竹体为字体样张,通过预处理、文 本行字切分、特征选择和分类识别的初步研究,获得对5种字体文本的平均识别率为89.582 % ,对其他字体的文本平均识别率为93.867%。
关键词:
印刷体藏文,
预处理,
特征选择,
分类识别
Abstract: The method of printed Tibetn character recognition presented by this paper is based on printed Tibetn character propectus of five different fonts,w h ich are Beiti,Heiti,Yuanti,Changti,Zhuti. Pilot study by preprocessing, docume nt line and character segmentation ,features selection and pattern classificati on, has obtained average recognition rate of 89.582% for doucments of those fiv e fo nts ,and average recognition rate for doucments of others fonts has reache d 93. 867%.
中图分类号:
王维兰,丁晓青,陈力,王华. 印刷体现代藏文识别研究[J]. , 2003, 29(03): 41-.
WANG Weilan1,DING Xiaoqing2,CHEN Li2,WANG Hua2. Study on Printed Tibetn Character Recognition[J]. , 2003, 29(03): 41-.