摘要
在金属标牌字符的识别过程中,预处理后的字符经常存在着断裂的现象,这给下一步的特征提取带来了困难。提出了一种新的修补方法,该方法利用字符本身的结构特征和预先计算的字符有关信息的统计值,有效的修补了断点。实验表明,与传统的修补方法相比,断裂字符利用本文算法修复后,字体边缘光滑,结构完整,满足特征提取的要求。
In the process of the recognition of metal label characters, the characters often have breakpoints,which brings troubles to the next process. A novel method is developed based on the characters＇ structural features and the previous statistic information,and this method mends these breakpoints effectively. As the experiments show, the broken characters mended have smooth edge and complete structure,which meets the needs of the next feature extraction.
出处
《光电子．激光》
EI
CAS
CSCD
北大核心
2008年第5期692-695,共4页
Journal of Optoelectronics·Laser
基金
教育部博士点基金资助项目(20060422011)
山东省2001年产业化重点项目资助(0203c06)
关键词
断裂修复
字符修补
曲线拟合
breakpoints connection
characters mending
curved-line fitting
