基于极性复指数变换的图像伪造检测算法  

Image forgery detection algorithm based on polarity complex exponential transform

作　　者：苏百兖[1] 杜永生[2] 黄传波[3] SU Baiyan;DU Yongsheng;HUANG Chuanbo(Department of Electronic Information Engineering,Jining Polytechnic of China,Jining Shandong 272073,China;School of Computer Science,Jining College,Qufu Shandong 273155,China;School of Defense Technology,Southwest University of Science and Technology,Mianyang Sichuan 621010,China)

机构地区：[1]济宁职业技术学院电子信息工程系,山东济宁272073 [2]济宁学院计算机科学学院,山东曲阜273155 [3]西南科技大学国防科技学院,四川绵阳621010

出　　处：《太赫兹科学与电子信息学报》2021年第3期497-505,共9页Journal of Terahertz Science and Electronic Information Technology

基　　金：国家自然科学基金资助项目(90820306);山东省科技发展计划资助项目(2014GGX109002)。

摘　　要：为解决当前图像伪造检测方法在识别复制内容区域时忽略了颜色信息和不同颜色分量之间的相关性,使其对伪造内容的定位与检测准确度不理想的问题,设计了基于改进的加速稳健特征(SURF)描述符与多元极性复指数变换的图像伪造检测算法。引入高斯低通滤波器,对彩色图像完成过滤,以消除噪声,再计算滤波图像的颜色不变性,用其替代SURF描述符中的灰度分量,对SURF方法予以改进,获取新的Hessian矩阵,充分检测彩色图像中的兴趣点;随后,利用这些兴趣点来构建一组连通的Delaunay三角网。基于四元极性复指数变换,充分考虑不同颜色分量之间的相关性,有效提取三角网的局部视觉特征;计算视觉特征之间的欧式距离,根据预设阈值,对三角网实施配准;最后,引入随机样本一致性,剔除错误匹配的三角网,并定义后处理方法,检测出复制伪造区域。测试数据显示:相对已有的复制-粘贴伪造检测方法,在多种几何变换条件下,所提方法具有更高的伪造检测准确性。In order to solve the problem as low location and detection accuracy of duplicate content areas induced by ignoring the color information and the correlation between different color components in current image forgery detection method when identifying the duplicate content area,this paper designs an image forgery detection method based on improved Speeded Up Robust Features(SURF)descriptor and multi-polarity complex exponential transformation.The Gaussian low pass filter is introduced to filter the color image for eliminating noise.Then the color invariance of the filtered image is calculated,and the gray component of the SURF descriptor is replaced by the color invariance to improve the SURF method for obtaining a new Hessian matrix and detecting the interest points in the color image adequately.Subsequently,a set of connected Delaney triangular networks is constructed by using these interest points.The local visual features of triangular networks are extracted based on the quadripolar complex exponential transformation.The Euclidean distance between visual features is calculated,and the triangulation mesh is registered according to the preset threshold.Finally,random sample consistency is introduced to eliminate mismatched triangular networks,and post-processing method is defined to locate the duplicated and forged regions.The test data show that compared with the existing copy-paste forgery detection methods,the proposed method has higher accuracy in forgery detection under various geometric transformations.

关 键 词：图像伪造检测 颜色不变性 SURF描述符 兴趣点 复指数变换 局部视觉特征 随机样本一致性 

分 类 号：TN911.73[电子电信—通信与信息系统]