基于Radial Shifted Legendre矩的图像取证算法  

Image tampering forensics method based on radial shifted Legendre moments

作　　者：赵杰 李营博 贺伟朝 Zhao Jie;Li Yingbo;He Weizhao(School of Electronic Information and Electrical Engineering,Shangluo University,Shangluo 726000,China)

机构地区：[1]商洛学院电子信息与电气工程学院,商洛726000

出　　处：《电子测量技术》2020年第13期85-88,共4页Electronic Measurement Technology

基　　金：国家自然科学基金(61806032);陕西省大学生创新创业训练计划项目(S201911396007);陕西省教育厅专项科研计划(17JK0238)项目资助。

摘　　要：Copy-Move篡改之后附加后操作攻击成为当前研究的难点和热点。为提高检测成功率,提出一种基于图像局部Radial Shifted Legendre矩的Copy-Move篡改检测算法。将检测图像分为若干相互重叠的区域块,提取各图像区域块的Radial Shifted Legendre矩作为该块的特征向量。利用各块的特征向量结合分块的位移偏移量进行分块相似性匹配,初步定位检测区域,再去除孤立的小匹配块,得到最终检测结果。实验仿真结果表明,该方法能够检测出Copy-Move篡改,并且相对传统DCT方案和局部统计特征方法,该算法对模糊、对比度变换、JPEG压缩、较小幅度旋转等后处理具有较好鲁棒性。Post operation attack after copy move tampering is the difficulty and hotspot of current research. In order to improve the correct detection rate, detecting algorithm based on image Legendre moments was proposed. The detected image was divided into overlapping blocks. Then Shifted Legendre moments of each block were obtained as feature vector. Feature vectors of each block and displacement offset were used to implement block similarity matching and preliminary detection results are obtained. Image morphological operation was employed to get the final test results by removing the mismatched blocks. Simulation results indicated that this method can detect Copy-Move tampering and is more robust to post-processing such as blur, contrast transformation, JPEG compression, small amplitude rotation than traditional DCT scheme and local statistical feature method.

关 键 词：篡改 LEGENDRE矩 图像分块 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]