基于特征像素相关性的虚假图像盲检测算法

Blind Detection of Digital Forgeries Using Correlation of the feature pixel

  • 摘要: 复制粘贴型(Copy-Move)篡改方式是把数字图像中一部分区域进行复制并粘贴到同一幅图像的另一个区域,以达到去除图像中某一重要内容的目的,是一种简单而有效的图像篡改技术。本文针对这种篡改方式,提出了一种利用图像特征像素间的相关性检测篡改图像的盲检测方法,为去除图像中少量噪声,降低算法的复杂度,本文对图像进行双树复小波变换(DTCWT),之后对图像块进行最大类间方差阈值计算和二值化处理,最后计算各个图像块的相关性。实验结果表明,本文提出的篡改认证方法能有效地对抗噪声对认证的影响,对一般的噪声(高斯噪声、低频滤波)具有很好的鲁棒性。

     

    Abstract: Copy and paste-type form (Copy-Move) of tampering with the digital image is that copies the part of the region and pastes it into another area with an image. This way is in order to achieve the removal of the important contents of an image. It is a simple and effective technology of the image tampering. In response to this distorted way, this paper presents a blind detection method that detects the correlation of the image feature pixel. In order to remove the little noises and reduce the computational complexity, this paper also use the DTCWT accuracy in the image, then calculates the threshold of the image blocks, the last calculate the correlation of the image blocks. The experimental result indicates that tampering authentication method proposed in this paper can effectively combat the impact of noise on the certification; this approach has the good robust to the general noise such as Gaussian noise and low frequency filter.

     

/

返回文章
返回