基于LBP和纹理信息挖掘的打印文件机源认定
发布时间：2019-07-13 17:09
【摘要】：为了消除光照对打印文件图像采集的影响,采用均匀模式的LBP对采集到的字符图像进行处理,提取LBP特征值,使用SVM进行打印文件的分类鉴别,分类准确率只有83.40%,说明LBP特征可以用于打印文件的分类鉴别,但是分类准确率不能满足实际应用的需要。运用两因子方差分析模型得出字符因子和打印机因子对鉴别率的影响都是显著的。采用纹理信息挖掘的方法消除字符因子的影响,研究打印机因子对打印文件鉴别的影响,可以将打印文件鉴别率提高到94.50%,具备实际应用的能力。
[Abstract]:In order to eliminate the influence of light on printing file image acquisition, uniform mode LBP is used to process the collected character image, LBP eigenvalues are extracted, and SVM is used for classification and identification of print files. The classification accuracy is only 83.40%, indicating that LBP features can be used for classification and identification of printing files, but the classification accuracy can not meet the needs of practical applications. Using the two-factor variance analysis model, it is concluded that the influence of character factor and printer factor on the discrimination rate is significant. By using the method of texture information mining to eliminate the influence of character factor and to study the influence of printer factor on the identification of printed file, the identification rate of printed file can be increased to 94.50%, which has the ability of practical application.
【作者单位】： 武汉大学电子信息学院;
【基金】：文件检验鉴定公安部重点实验室(中国刑事警察学院)项目(11KFKT001)
【分类号】：TP334.8;TP391.41
，
本文编号：2514154
[Abstract]:In order to eliminate the influence of light on printing file image acquisition, uniform mode LBP is used to process the collected character image, LBP eigenvalues are extracted, and SVM is used for classification and identification of print files. The classification accuracy is only 83.40%, indicating that LBP features can be used for classification and identification of printing files, but the classification accuracy can not meet the needs of practical applications. Using the two-factor variance analysis model, it is concluded that the influence of character factor and printer factor on the discrimination rate is significant. By using the method of texture information mining to eliminate the influence of character factor and to study the influence of printer factor on the identification of printed file, the identification rate of printed file can be increased to 94.50%, which has the ability of practical application.
【作者单位】： 武汉大学电子信息学院;
【基金】：文件检验鉴定公安部重点实验室(中国刑事警察学院)项目(11KFKT001)
【分类号】：TP334.8;TP391.41
，
本文编号：2514154
本文链接：https://www.wllwen.com/kejilunwen/jisuanjikexuelunwen/2514154.html
