【作者】 吉建华；

【导师】 张静；

【作者基本信息】 天津大学 ， 信号与信息处理， 2008， 硕士

【摘要】 近年来,随着计算机网络通信技术的飞速发展,各种形式的数字媒体得到了广泛的应用。但是,由于功能强大的图像处理软件的广泛使用,数字图像越来越容易被篡改。传统上,在鉴别图像真伪时,采用主动检测的方法,即:事先在数字图像中嵌入数字水印或签名的方法。但是,该方法存在着诸多的缺陷和不足。因此,仅根据图像自身进行图像真实性和完整性认证的被动检测方法成为了新兴的热门研究方向。在通过对图像进行拷贝、粘贴等操作而达到图像篡改目的的过程中,往往会在图像篡改部分与原图像的接合处出现瑕疵(奇异)点。而小波变换系数模的局部极大值与图像的奇异点(突变点)有着特定的联系,因此,可以据此实现对图像的篡改检测。本论文根据小波系数的规则性,提出了一种JPEG2000图像篡改的被动检测方法。其算法原理是:小波系数的模取值越大,那么,小波变换尺度和小波系数模的局部极大值的对数就越近似呈线性关系。该方法首先提取图像的边缘信息,然后采用二进小波变换获取图像各子带中对应像素点的信息,最后对小波系数模的局部极大值的对数进行一维曲线拟合,从而检测图像是否经过篡改,并对篡改部分进行定位。实验结果表明,本算法对JPEG2000图像有较好的检测效果,可以较为准确地确定篡改部位。另外,本算法检测效率较高,运行时间较短,因此,它是一种效率和效果兼顾的数字图像被动检测算法。更多还原

【Abstract】 Nowadays, with the rapid development of computer network and communication technology, various forms of digital media have been widely used. However, due to the widespread use of powerful image-processing software, it is increasingly easier to tamper digital images. The traditional method to distinguish trueness or fake is to embed digital watermarks or signatures in digital images in advance. But it still has a lot of weaknesses. Therefore it forms a new and important research direction to authenticate digital images without watermarks or signatures, which is called passive image detection.In the course of image forgery with paste-copy, forgers often produce some singularity points at the edge of image junction. A specific contact between the logarithm of local maximum value of wavelet transform coefficient and the singularity points of image can be used to achieve the tamper detection of images.This dissertation presents a new JPEG2000 image tamper detection measure based on regularity of wavelet coefficients. Its algorithm principle is: Firstly, this method extracts the image edge information. Then it accesses to the corresponding pixels of information on the image sub-bands with UDWT (Undecimated Discrete Wavelet Transform). Finally, curve fitting is used. The experimental results show that the algorithm can detect JPEG 2000 tampered image ideally. In addition, this method can be used effectively.更多还原