基于U2-Net深度网络和投影波谷法的在架图书错序检测  

On⁃shelf book disorderly detection based on U2⁃Net deep network and project valley⁃fold

作　　者：王红芳 王宇航 宣静雯 张凯兵 WANG Hongfang;WANG Yuhang;XUAN Jingwen;ZHANG Kaibing(Library of Xi’an Polytechnic University,Xi’an 710048,China;School of Electronics and Information,Xi’an Polytechnic University,Xi’an 710048,China;School of Computer Science,Xi’an Polytechnic University,Xi’an 710048,China)

机构地区：[1]西安工程大学图书馆,陕西西安710048 [2]西安工程大学电子信息学院,陕西西安710048 [3]西安工程大学计算机科学学院,陕西西安710048

出　　处：《现代电子技术》2024年第3期149-154,共6页Modern Electronics Technique

基　　金：陕西省自然科学基础研究计划重点项目(2018JZ6002);西安工程大学哲学社会科学研究项目:知识产权信息服务在成果转化中的应用研究(2019ZXSK13);陕西省图书馆学会研究课题B类课题(231042)。

摘　　要：针对在架图书错序检测中图书索书号区域的准确定位关键问题,提出一种基于U2⁃Net深度网络和投影波谷法的在架图书索书号区域定位识别方法。该方法首先使用U2⁃Net深度网络对在架图书索书号区域进行预分割,然后通过投影波谷法对每本图书的索书号区域进一步精确分割,最后对分割出的索书号区域图像使用光学字符识别技术,完成对索书号的识别,通过对识别结果进行排序关系判断,完成错序检测过程。实验结果表明,该方法能实现准确的索书号区域图像分割,为图书馆智能化管理提供了有效的解决方案。In order to address the key issue of accurate localization of book index areas in on⁃shelf book disorderly detection,an on⁃shelf book index area localization and recognition method based on the U2⁃Net deep network and the project valley⁃fold method is proposed.In this method,the U2⁃Net deep network is employed for pre⁃segmentation of the on⁃shelf book index areas first,and then the project valley⁃fold method is utilized to further precisely segment the index areas of each book.Subsequently,optical character recognition(OCR)technology is applied to the segmented index area images to accurately recognize the book index numbers.Finally,the disorderly detection is completed by judging the sorting relationship based on the recognized results.Experimental results demonstrate that the proposed method can achieve accurate segmentation of the book index area images,so it can provide an effective solution for intelligent library management.

关 键 词：深度网络 U2⁃Net 投影波谷法 图书错序检测 光学字符识别 索书号定位 

分 类 号：TN911.73-34[电子电信—通信与信息系统] G258.0[电子电信—信息与通信工程]