赵洁. 联合Harris特征点与Hu不变矩的图像区域复制篡改盲取证[J]. 微电子学与计算机, 2013, 30(11): 15-17,22.
引用本文: 赵洁. 联合Harris特征点与Hu不变矩的图像区域复制篡改盲取证[J]. 微电子学与计算机, 2013, 30(11): 15-17,22.
ZHAO Jie. Blind Forensics for Image Region-Duplication Forgery Using Harris Points and Hu Invariant Moments[J]. Microelectronics & Computer, 2013, 30(11): 15-17,22.
Citation: ZHAO Jie. Blind Forensics for Image Region-Duplication Forgery Using Harris Points and Hu Invariant Moments[J]. Microelectronics & Computer, 2013, 30(11): 15-17,22.

联合Harris特征点与Hu不变矩的图像区域复制篡改盲取证

Blind Forensics for Image Region-Duplication Forgery Using Harris Points and Hu Invariant Moments

  • 摘要: 提出了一种将 Harris特征点和 Hu不变矩相结合的图像区域复制篡改检测方法。首先采用 Harris算子检测图像中的角点作为特征点,然后以特征点为中心计算其邻域内Hu矩的前4维低阶矩形成特征向量,最后记录相等位移矢量的发生频率并通过阈值化处理得到匹配的特征点,从而标识复制粘贴区域。仿真实验表明,该算法不仅可以有效检测多区域复制粘贴篡改操作,而且能够有效抵抗多种篡改后处理操作,包括加性高斯白噪声,JPEG压缩,对比度、亮度和曝光度调整以及JPEG压缩和加噪的混合操作。

     

    Abstract: A detection method of image region-duplication forgery is proposed using Harris points and Hu invariant moments.First,Harris operator is applied to detect corners as feature points,then the former four dimentions of Hu moments are computed in the neighborhood of which as the center to constitute feature vector,and finally the occurrence frequency of position transfer vector is recorded to obtain matching interest points with thresholding processing,thus copy-move regions are marked.Experimental results show that the algorithm can not only detect multi-region copy-move manipulation effectively, but also be able to effectively resist various tampering post-processing operations, including additive white gaussian noise, JPEG compression, contrast, brightness and exposure adjustment and the hybrid operation of JPEG compression and noise addition.

     

/

返回文章
返回