视频与图像信息隐藏技术若干问题研究

发布时间：2018-05-11 12:30

本文选题：信息隐藏 + 数字水印　；参考：《中国科学技术大学》2017年硕士论文

【摘要】：随着互联网的传输能力和计算机信息处理能力的提升,多媒体信息已经可以方便高效地在互联网中传播,并可以让人们任意地拷贝与修改。在此趋势下,多媒体信息的内容安全性已经成为近20年来人们普遍关注的问题。在众多的信息安全技术中,加密和信息隐藏技术成为了多媒体信息内容隐私保护、篡改检测、版权管理等等的主要手段。如果按照嵌入域分类,信息隐藏技术可以分为空域(或时域)和变换域方法,其中变换域方法包括了近几年被提出的加密域方法,如果按照可逆性分类,又可以分为可逆信息隐藏和不可逆的信息隐藏。本文根据信息隐藏的不同类别,分别研究了图像和视频的空域以及加密域、压缩域信息隐藏,并着重研究了信息隐藏技术中的可逆性和容量-失真性能。本文所研究的内容主要分三个部分,分别为空域的可逆信息隐藏、DCT域水印算法及其优化、以及视频加密域可逆水印。这三部分分别对应空域、变换域、加密域方法。本文的主要贡献有以下三点:1、提出一种基于像素灰度值预测差值的直方图平移方案。首先采用Warped Distance算法来做像素值的预测,再在此基础上引入图像的局部梯度以实现对像素值更精确的预测,综合上述两种策略改进了现有的直方图平移算法。实验结果表明该方案相对于近年来的其他方案能够在保证图像质量的情况下同时有效提升数据嵌入量。2、利用一种基于结构相似性(Structural Similarity Index,SSIM)和动态规划算法的率失真优化策略在JPEG压缩的同时嵌入最优水印,试验数据表明该方案提高了含水印图像的率失真性能。3、提出一种基于H.265/HEVC视频编码标准的加密域可逆数字水印方案,并利用它实现视频的认证加密。同时实现内容安全性保护与完整性认证。实验表明该方案不仅能在加密的视频中嵌入可逆水印,还能对视频的篡改进行定位。相对于近几年的其他方案,本文的方案提高了信息隐藏的容量-失真性能,并实现了信息隐藏的可逆性,加密域、压缩域嵌入以及率失真优化,也给今后其他的相关研究提供了一些可行的思路。
[Abstract]:With the improvement of the transmission ability of the Internet and the computer information processing ability, multimedia information can be easily and efficiently spread in the Internet, and people can copy and modify it arbitrarily. In this trend, the content security of multimedia information has become a common concern in the past 20 years. Among many information security technologies, encryption and information hiding technology have become the main means of multimedia information privacy protection, tampering detection, copyright management and so on. If classified according to embedded domain, information hiding technology can be divided into spatial domain (or time domain) and transform domain method, where transform domain method includes encryption domain method proposed in recent years, if classified according to reversibility, It can be divided into reversible information hiding and irreversible information hiding. According to the different categories of information hiding, this paper studies the spatial domain of image and video, the encryption domain and the compressed domain of information hiding, and focuses on the reversibility and capacity-distortion performance of information hiding technology. The content of this paper is divided into three parts: reversible information hiding in spatial domain and its optimization in DCT domain, and reversible watermarking in video encryption domain. These three parts correspond to spatial domain, transform domain and encryption domain method respectively. The main contributions of this paper are as follows: 1. A histogram translation scheme based on pixel gray prediction difference is proposed. First, the Warped Distance algorithm is used to predict the pixel value, and then the local gradient of the image is introduced to achieve more accurate prediction of the pixel value. Based on the above two strategies, the existing histogram translation algorithm is improved. The experimental results show that the proposed scheme can effectively improve the embedding amount of data at the same time, compared with other schemes in recent years, and make use of a structural Similarity IndexSIMI based on structural similarity and the rate distortion of dynamic programming algorithm. The optimal watermark is embedded in the JPEG compression strategy at the same time. The experimental data show that the proposed scheme improves the rate-distortion performance of the image with watermark. A reversible digital watermarking scheme in encryption domain based on H.265/HEVC video coding standard is proposed and used to realize video authentication and encryption. At the same time, the content security protection and integrity authentication are realized. Experiments show that this scheme can not only embed reversible watermark in encrypted video, but also locate the tampering of video. Compared with other schemes in recent years, this scheme improves the capacity distortion performance of information hiding, and realizes the reversibility of information hiding, encryption domain, compression domain embedding and rate distortion optimization. It also provides some feasible ideas for other related research in the future.
【学位授予单位】：中国科学技术大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP309

【参考文献】