摘要
综合运用图像预处理、图像分割、模板匹配等算法,提出了一种基于连通域特征的射线胶片字符识别技术,并在MATLAB下实现。首先对图像进行滤波、二值化和图像分割等预处理,得到待识别的字符区域,然后在该区域内对待识别的字符进行连通域处理,分析分割后的字符相对于字符模板的连通域特征,再进行模板匹配并采用最大相似法实现字符识别。试验结果表明该技术能够有效提取出数字化胶片中的字符信息,并具有较好的鲁棒性。
The integrated use of algorithms of image preprocessing,image segmentation, template matching, a ray-film character recogni tion technology is proposed based on connected domain characteristic and achieved in MATLAB. First, the image preprocessing is con ducted,including image filtering, binarization and segmentation, get the character area to be identified. Then used connected domain pro- cessing method in the region for the characters to be identified, and analyzed connected domain characteristics of the character after the split in relation to the character templates. Finally, the template is matched and achieve character recognition using maximum likelihood method. The results show that the technique can effectively extract information out of the film characters, and has better robustness.
出处
《计算机技术与发展》
2012年第12期119-122,共4页
Computer Technology and Development
基金
陕西省教育专项科研计划资助项目(11JK0921)
山东省科技发展计划项目(2010GWZ20402)
质检工艺性行业科研专项(201010030)
山东省质检总局科研项目(2009KYZ22)
关键词
连通域
字符分割
字符识别
模板匹配
数字化胶片
connected domain
character segmentation
character recognition
template matching
digital Iilm