摘要
为了解决当前图像伪造检测算法主要是在图像空域中定位伪造区域,难以降低图像维数,使其复杂度大;且不能有效检测几何变换篡改形式的伪造区域,导致其鲁棒性不佳的不足,提出了离散小波变换耦合静电场理论的图像伪造检测算法;首先,引入离散小波变换,提取伪造图像的低频子带,降低图像空间;再基于静电场理论,将提取子带映射到虚拟电场中,提取鲁棒性较强的特征,利用Radix排序算法对特征完成重组,形成特征矩阵;最后,定义相同仿射变换,并用其处理排序矩阵,完成伪造区域检测;实验测试结果显示:与当前的移动复制伪造检测技术相比,所提算法具有更高的定位效率与检测精度;同时拥有较强的鲁棒性,有效抗击几何变换篡改;该算法能够高效精确检测几何变换伪造形式的图像内容。
In order to solve the problems such as big complexity induced by difficultly reducing the image dimension mainly in the space domain for locating forged regions,and the poor robustness caused by difficultly detecting geometric transform tamper form the forged areas in current image forgery detection algorithm,the image fast forgery detection algorithm is proposed based on discrete wavelet transform coupled electrostatic field theory.Firstly,the low frequency subbands of forged images were extracted by introducing the discrete wavelet transform for reducing the image space.Then the strong robustness feature was extracted based on electrostatic field theory for mapping the extracted subbands to virtual electric field,as well as these features were ordered by Raster scanning to form feature matrix.Finally,the affine transformation was used to deal with the matrix for finishing the forgery detection.Test results show that：this algorithm had higher location efficiency and detection precision,as well as strong robustness to effectively against geometric transform tampering compared with the current move-copy forgery detection technology.This algorithm can efficiently and accurately detect the image content of forgery image with geometric transformation.
出处
《计算机测量与控制》
2016年第3期44-47,共4页
Computer Measurement &Control
基金
四川省教育厅理工科重点项目(14ZA0339)
关键词
图像伪造检测
离散小波变换
静电场
Radix排序
仿射变换
image forgery detection
discrete wavelet transform
electrostatic field
raster scan
affine transformation
