云计算环境中基于水印的多媒体认证技术研究

Cloud computing represents the trends of technical development because of its low cost, fast and almost unlimited scalability, and the security concerns of user have become the biggest obstacle to the application. The defect of classical cryptography applied in multimedia authentication makes the problem more prominent. Given this background, the project research on several key issues of multimedia origin and content authentication based on digital watermarking. The main contents include: Research of segmentation of interested content, feature extraction, quantification and watermark generation; Research of perceptual hashing based on the human sensory characteristics;Study the watermark formation mechanism to anti framed and deny as well as the parameters optimization of cloud model based watermarking droplet generating; Research of novel robust watermarking algorithm which have efficiency priority and its speed optimization; Design of novel content based semi-fragile watermarking algorithm and its safety testing; According to the advantage of cloud computing environment, security protocols of watermark embedding, detecting and arbitrating are designed and proved; Development of cloud middleware to color multimedia data and authenticate content. Implementation of prototype system which integrated relevant research results. The multimedia in project here refers to the image, video as the main research object, content tamper detection which is consistent with human subjective sensation as a key technology. Considering application background in cloud computing environment, the research goal is clear and the technical route is feasible in the project.

云计算因其低廉、快速以及近乎无限的伸缩性代表了技术发展的趋势，而用户对安全的疑虑已成为应用向云计算迁移的最大障碍。经典密码学在多媒体认证领域的不足则使问题更为突出。考虑此背景，本项目基于数字水印，就多媒体溯源认证和内容认证所涉及的若干关键问题展开研究。主要研究工作包括：感兴趣内容的分割、特征提取、量化及水印生成研究；基于人类感觉特性的多媒体感知哈希研究；防诬陷和抵赖的水印形成机制及基于云模型的水印云滴参数优化；效率优先的新型鲁棒水印算法设计及其速度优化研究；内容相关的新型半脆弱水印算法设计及其安全性测试；充分利用云计算环境优势的水印嵌入、检测和仲裁协议及其形式化证明；多媒体数据标记和内容认证的云中间件开发；体现相关研究成果的原型系统实现。项目中所指的多媒体以图像、视频为主要研究对象，以与人眼主观感觉相一致的内容篡改检测为技术关键。考虑云计算环境应用实际背景，项目研究目标明确，技术路线可行。

本项目为小额资助项目,项目周期及研究内容变动较大,项目组在研究中仍以多媒体的认证为目标,循序展开研究。主要研究工作和成果包括：对现有Hash码生成算法的分析及成功欺骗;图像内容的特征表达及基于尺度不变特征的水印算法研究; 基于混沌序列的新型图像加密算法及其安全性验证;基于GPU的加速研究及算法实现。