抗翻转、旋转和缩放攻击的图像区域复制篡改检测.pdfVIP

第27 卷 第 1 期 计算机辅助设计与图形学学报 Vol. 27 No.1 2015 年 1 月 Journal of Computer-Aided Design Computer Graphics Jan. 2015 抗翻转、旋转和缩放攻击的图像区域复制篡改检测 邵 虹, 朱 虹, 崔文成* (沈阳工业大学信息科学与工程学院 沈阳 110870) (shaoh1974@) 摘 要: 针对现有图像区域复制篡改检测算法对几何攻击较脆弱的特点, 提出一种能够有效抵抗翻转、旋转和缩放 的篡改检测方法. 首先采用对数极坐标变换将圆窗图像投影到具有旋转、缩放不变性的坐标空间; 然后自适应调整圆 窗大小与频带宽度, 消除多余图像信息与噪声的干扰, 提高相位相关峰值以实现图像圆窗匹配; 最后采用自适应调 整得到的圆窗图像计算相位相关矩阵峰值位置以获取旋转角度与缩放因子, 并将其应用于区域搜索, 最终实现被复 制区域与篡改区域的定位. 实验结果表明, 该方法对常用的几何变换操作准确率高、误检率低. 关键词：篡改检测; 区域复制; 几何攻击; 对数极坐标变换; 相位相关 中图法分类号：TP391 Detection of Image Region-Duplication Forgery Affected by Flipping, Rotation and Scaling * Shao Hong, Zhu Hong, and Cui Wencheng (School of Information Science and Technology, Shenyang University of Technology, Shenyang 110870) Abstract: Most existing region-duplication forgery detection algorithms are vulnerable to geometric distor- tions. Therefore, an effective tampering detection algorithm is proposed for resisting to geometric distortions which involves flipping, rotation, and scaling. Firstly, the circle window images are projected to the rotation and scaling invariance coordinate space by performing log-polar transform. Then circle window radius and frequency band width can be adjusted automatically, which can eliminate redundant image information and noise disturbance, and also increase the phase correlation peak in order to match the circle window images. Finally, the phase correlation can be calculated by adopting automatic adjustment circle window images to obtain the peak position, thus rotation angle and scaling factor can be got. These offsets are applied in region searching, and copied regions and tampered regions can be located. Experim