数字图像的完整性验证与自修复水印算法研究

【作者】 何孝富；

【导师】 张功镀； 黄继风；

【作者基本信息】 上海师范大学 ， 计算机应用技术， 2004， 硕士

【摘要】 随着多媒体和网络技术的迅速发展，对数字媒体的完整性、媒体内容的真实性认证显得日益重要。目前，图像编辑软件已经广泛普及，如Adobe PhotoShop、Microsoft Paint、PaintShop Pro等，利用这些软件很容易对一幅数字图像进行篡改，使得对图像内容的完整性提出质疑。本文的研究目的是利用水印技术对一幅数字图像的完整性进行验证，如果发现图像被篡改，应对篡改区域进行定位，并对篡改区域尽可能的修复。 目前用于数字图像完整性验证及自修复的数字水印方案主要有完全脆弱性水印、半脆弱性水印、自嵌入水印等，本文主要研究自嵌入及半脆弱数字水印，分别从空域及变换域角度详细分析当前国内外现有的各种算法，指出其不足之处，并提出了相应的改进算法，实验结果表明本文算法的有效性与实用性。本文的创新之处总结如下： (1)提出了一种改进的具有修复功能的空域自嵌入水印算法，利用分组编码保存量化后的DCT系数，改善了修复效果，克服了以往使用位图矩阵的不足，具有通用性与实用性。 (2)鉴于分形图像压缩具有很高的压缩比且重构图像效果好的优点，本文利用分形压缩实现了一种空域自嵌入水印算法，不但能对篡改区域进行有效定位，而且具有修复功能。 (3) 由于现有的空域自嵌入水印算法大多数是对位图图像嵌入水印，在抗JPEG压缩方面显得过于脆弱，因此本文实现了一种直接对JPEG图像进行自嵌入的数字水印算法，具有检测篡改及修复功能，对于当今网络时代来讲具有重要意义。 (4) 由于小波变换具有良好的空间—频率局部特性以及与人眼视觉特性相符的变换机制，在新一代静止图像压缩标准(JPEG 2000)和运动图像压缩标准MPEG-4中占据了重要位置。因此，本文利用小波系数均值来嵌入水印，提出了一种基于小波域的图像完整性验证的半脆弱水印算法，在鲁棒性和脆弱性方面达到了很好的平衡，对非恶意攻击有很强的鲁棒性，而对于恶意的篡改操作有很强的识别和定位能力。更多还原

【Abstract】 With the development of internet and multimedia technology, the authentication of digital media becomes more and more important. Powerful publicly available image processing software packages such as Adobe PhotoShop, Microsoft Paint and PaintShop Pro make digital forgery a reality. It is possible to carefully tamper an digital image while leaving barely detectable traces by using these tools. So, the problem is how to ensure the integrity of image content. The purpose of this thesis is to authenticate the integrity of a digital image using watermarking technology, which is to check if the image has been tampered and identify the tampered portions. If possible try to recover portions of the image that have been tampered without accessing the original image.In the past, several techniques and concepts based on data hiding or steganography have been introduced as a means for tamper detection in digital images and for image authentication-fragile watermarks, semi-fragile watermarks, and self-embedding watermarks. The research of this thesis mainly includes self-embedding and semi-fragile watermarks. We propose several watermarking techniques after analyzing other watermarking methods detailedly. The main contributions can be summarized as following:(1) Based on previous work, we proposed an improved spatial self-embedding watermarking technique. We save the quantized DCT coefficients using group encoding while not using bit matrix which restricts saving bit length of each DCT coefficient. This method has many advantages compared to the latter and is testified by many images.(2) Because fractal coding has the characteristic of high compression ratio, this thesis designed a spatial self-embedding watermarking technique. After self-embedding, it can detect any changes to a watermarked image as well as localizing the areas that have been tampered. It is also possible to recover portions of the image which have been tampered. Experiments on real images verify the effectiveness of the proposed technique.(3) Since the current self-embedding watermarking techniques are too fragile to resist noise or JPEG compression, we proposed a self-embedding method based on JPEG images which plays an important role in current internet environment.(4) Designed an image authentication method of semi-fragile watermarking technique based on DWT. We transform the image into wavelet domain and group the four adjacent wavelet coefficients. Utilizing the characteristics of the human visual system, we embed a digital signal into the average value of the four adjacent wavelet coefficients since the average value has better stability than single wavelet coefficient. This method needn’t original image when the watermark is extracting. Experiment results show the effectiveness of this method which is robust to common image process and fragile to malicious attack, i.e. this method can detect the tampered portions.更多还原