摘要
针对药瓶标签的特点,对字符定位、字符分割、字符识别等多个环节进行了深入研究,设计了一种字符识别系统。对药瓶标签图像进行预处理,为后续药品标签上字符的定位做准备。用轮廓跟踪法定位字符所在的大致区域,采用水平投影结合图像分割方法精确定位字符区域。利用垂直投影法定位字符中的粘连块,并进一步利用滴水算法分割粘连字符。选用K-L变换对粗分类后的子类提取字符特征,选用一种自适应调节学习率和动态调整S型激励函数相结合的改进BP算法对字符进行精确识别。实验证明本系统能够快速、高效地识别出药瓶标签上的字符,有实用价值。
Character recognition on medicine bottle label containing batch number, produc- tion date and expiry date is studied in this paper. It mainly consists of character location, character segmentation and character recognition. Firstly, it preprocesses the image of medi- cine bottle label for the following procedures;secondly, general character region is obtained by contour tracing and precise location is realized by horizontal projection algorithm;then it locates character adhesion region by vertical projection algorithm and segments the merged characters by drop fall algorithm; finally, it adopts Karhunen-Loeve transform to extract character features from the subclasses of coarse classification and accurately recognizes char- acters by the improved BP algorithm. Experimental results show that the system can recog- nize the bottle label characters quickly and accurately, which is of great practical value.
出处
《沈阳理工大学学报》
CAS
2014年第1期1-7,共7页
Journal of Shenyang Ligong University
关键词
药瓶标签
字符定位
字符分割
字符识别
BP神经网络
image processing
character location
character segmentation
character recogni- tion system
BP neural network