图书索书号识别算法及其实现  

Recognition Algorithm of Book Call Number and Its Implementation

作　　者：陈旭[1] 方建军[1] 刘彩霞[1] CHEN Xu FANG Jian-jun LIU Cai-xia(College of Automation, Beijing Union University, Beijing 100101 ,China)

机构地区：[1]北京联合大学自动化学院,北京100101

出　　处：《北京联合大学学报》2017年第2期66-71,共6页Journal of Beijing Union University

基　　金：北京市教委人才强教深化计划长城学者培育计划项目(CIT&TCD20150314);北京市自然科学基金项目(4142018)

摘　　要：针对多本在架图书图像中相邻图书的索书号分割、识别存在的识别率低的问题,提出了一种图书索书号识别算法,该算法实现了对单本图书索书号的分割、提取以及字符识别,具体做法是通过确定多阈值中的阈值范围来分割索书号区域,利用种子填充法将索书号区域中的字符提取出来,最后利用神经网络的方法完成对索书号字符的识别。实验结果表明,直立图书的字符识别率为95.3%,倾斜图书的字符识别率为93.1%,总的识别率为94.3%,达到了良好的识别率。Regarding the low recognition rate in the segmentation and recognition of book call number for adjacent books,this paper proposed a book number recognition algorithm, which realizes single book call number＇ s segmentation, extraction and character recognition. First, by determining the threshold value, the call number area is divided. Then, seed filling method is used to pick up the character of call number. Finally, the recognition of the characters is achieved using neural network. The experimental results show that the vertical book character recognition rate is 95.3%, the tilted book character recognition rate is 93.1%, and the overall recognition rate is 94.3%. Thus a good recognition rate is achieved.

关 键 词：索书号 识别算法 图像分割 图像提取 

分 类 号：TP391.43[自动化与计算机技术—计算机应用技术]