摘要
针对网络或新闻中出现的大量伪造、篡改图片,研究低质量有损压缩图片的篡改和伪造算法,以检测是否为篡改图片。本文以同幅图像的移动-复制模型为例,进行了同幅图像的篡改取证,利用图像本身的内在统计特性,以块特征为基础,基于字典排序流程,利用三种不同的盲取证算法:DCT变换,PCA变换,小波变换,检测图像中的篡改区域;最后,对比分析三种算法在降采样和加噪声处理下的算法鲁棒性,得到了较好的实验结果,对数字图像的盲取证算法进行了有力的探索。
Presently, there are large number of forged/tampered pictures showing on the Internet. Therefore, an algorithm is really required on images of low quality and forging/tampering so as to detect the authenticity of one image. With the copy-move manipulation to tamper an image as the example subject, the analysis was made of the image itself and its inner statistical properties based on the arrangement by a resultant dictionary that was from the matrix out of the block features inherent in the copy-move function. Three kinds of blind forensic arithmetic, DCT, PCA and WAVELET, were used to detect the image's tampered area. Lastly the algorithm's robustness was compared through the three kinds of arithmetic to receive sampling and noise adding, thereby getting a better experimental result.
作者
王云峰
韩常
Wang Yunfeng;Han Chang(Gansu Institute of Political Science and Law,Lanzhou 730070,China)
出处
《刑事技术》
2018年第3期222-225,共4页
Forensic Science and Technology
基金
甘肃省科技计划项目立项自然科学基金(No.145RJZA055)
甘肃省财政厅2012年度高校基本科研项目(甘财教[2012]129号)
甘肃政法学院重点项目(GZF2014XZDLW07)
关键词
图像
盲取证
图像取证
字典排序
images
blind forensics
image tampering
arrangement by a block-featured dictionary