打印文稿识别技术研究与设计  

Research and Design of Print Document Recognition Technology

作　　者：孟令军[1] 文波[2] 张晓春[2] 韩朝辉[2] 

机构地区：[1]中北大学电子测试技术国家重点实验室,太原030051 [2]中北大学仪器科学与动态测试教育部重点实验室,太原030051

出　　处：《科学技术与工程》2015年第14期185-190,195,共7页Science Technology and Engineering

摘　　要：近年来由于现代数字电子技术的飞速发展,各类合同、文档、传单的真实性有待于考证;而且也迫切需要准确掌握各类打印文档的来源。针对打印文档的真伪性问题,提出在对单个字符图像进行灰度级压缩和二值去噪的基础上,利用统计法中的灰度共生矩阵来提取字符图像的特征值;并以这些特征值建立相应的样本数据库,借助神经网络工具箱对特征值进行分类和识别。试验对试验条件的控制较为严格,样本采集为5种全新打印机打印的文档各取5张,字符图像的提取采用同一扫描仪,并且控制工作时间和打印文档的数目一致。实验结果表明,观察提取的特征数据发现,不同打印文档的特征数据之间有明显差别,而且特征数据的正确性和可靠性明显提高。试验中对5种打印机全部正确识别。In recent years,due to the rapid development of modern digital electronic technology,types of contracts,documents,leaflets authenticity to be verified,but also an urgent need for accurate information on the source of all kinds of printed documents.The authenticity of the document for printing problem,a single characterbased grayscale image compression and de-noising based on binary to extract characteristic value of the character images using statistical method GLCM be can used,and these eigenvalues establish the appropriate sample database,using neural network toolbox eigenvalues classification and identification.The trial of the test conditions more stringent controls,sample collection for the five kinds of new printer to print a document from each of five extracted character images using the same scanner,and control the number of working hours in line and print documents.Experimental results show that the extracted feature data observation found a significant difference between the different characteristics of the print document data,and the accuracy and reliability of the feature data has improved significantly.The test correctly identified all five kinds of printers.

关 键 词：打印文档取证 共生矩阵 分类器 神经网络 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]