基于深度学习的打印文档缺陷检测算法  

Print document defect detection algorithm based on deep learning

作　　者：刘李漫[1] 汪梦婷 劳喜鑫 吴兴宇 LIU Liman;WANG Mengting;LAO Xixin;WU Xingyu(Biomedical Engineering Collage, South-Central University for Nationalities, Wuhan 430074, China)

机构地区：[1]中南民族大学生物医学工程学院,武汉430074

出　　处：《中南民族大学学报（自然科学版）》2021年第5期504-511,共8页Journal of South-Central University for Nationalities：Natural Science Edition

基　　金：国家自然科学基金资助项目(61976227);湖北省自然科学基金资助项目(2019CFB622)。

摘　　要：在工业生产中,常常需要检测大量的打印文档.现有的打印文档缺陷检测通常采用基于图像处理的方法,该方法容易受到外界环境的干扰,且误差相对较大.为了解决这一问题,提出了一种基于深度学习的打印文档缺陷检测算法.该算法包括打印文档纸张缺陷检测、打印文本倾斜和偏移检测、打印文字清晰度检测三个部分.通过分类网络先检测纸张全局上的缺陷问题,再通过对比网络检测纸张局部细节上的缺陷问题.实验表明,所提出的算法不仅能够同时检测打印文档全局和细节上的缺陷,还能减少实验环境等外界因素的干扰,可取得令人满意的检测效果,具有较好的实用价值.It is often necessary to examine a large number of printed documents in industrial production.The existing defect detection method of printed documents is usually based on image processing.The method is easy to be disturbed by the external environment and the error is relatively large.In order to address this problem,a document defect detection algorithm based on deep learning is proposed.The algorithm includes three parts:paper defect detection of printed document,tilt and offset detection of printed text,and definition detection of printed text.The paper defects are firstly detected globally through the classification network,and then the paper defects are detected locally through the comparison network.The experimental results show that the proposed algorithm can not only detect the global and detailed defects of printed documents simultaneously,but also reduce the interference of external factors such as experimental environment,and achieve a satisfactory detection effect,so it has good practical value.

关 键 词：文档缺陷检测 分类网络 字符对比 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]