Perspectives on Information Hiding for Multimedia Security

多媒体安全信息隐藏的观点

• 批准号: 0242417
• 负责人: Rajarathnam Chandramouli
• 金额: $ 28万
• 依托单位: Stevens Institute of Technology
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2003
• 资助国家: 美国
• 起止时间: 2003-07-01 至 2007-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0242417&HistoricalAwards=false
• 关键词: Perspectives; Information; Hiding; Multimedia; Security

A fully digitized multimedia enriched world will be a reality when security and quality of the multimedia content can be guaranteed. Information hiding (watermarking, digital fingerprinting, etc.) has been gaining popularity in this quest for security. Investigation of information hiding based security techniques that suit multimedia data and networks is the goal of this project. Primarily, this project considers problems in providing multimedia security from the following different perspectives: (a) Encoder/Content originator perspective, (b) Decoder/Content recipient perspective, (c) Attacker's perspective, and (d) Networking perspective. Studying the multimedia data security problem from these perspectives could lead to a better understanding of security issues from an end-to-end system view point. Connections between perceptually based and information theory based information hiding, and content based watermarking will be investigated with robustness and embedding capacity being the main focus. Detector dependent embedding and game theoretic formulation to study the dynamics between the decoder and a malicious attacker are proposed. A new attack dependent definition of security and methods to compute the information hiding capacity based on this definition are also proposed in this project. Finally, joint source-channel coding and multiple description coding approaches for robust watermark systems that can handle network losses are also of interest.

当多媒体内容的安全性和质量得到保证时，完全数字化的多媒体世界将成为现实。在这种对安全性的追求中，信息隐藏（水印、数字指纹等）越来越受欢迎。该项目的目标是研究适合多媒体数据和网络的基于信息隐藏的安全技术。该项目主要从以下不同角度考虑提供多媒体安全性的问题：(a) 编码器/内容发起者角度，(b) 解码器/内容接收者角度，(c) 攻击者角度，以及(d) 网络角度。从这些角度研究多媒体数据安全问题可以从端到端系统的角度更好地理解安全问题。将研究基于感知和基于信息论的信息隐藏与基于内容的水印之间的联系，其中鲁棒性和嵌入能力是主要关注点。提出了检测器相关嵌入和博弈论公式来研究解码器和恶意攻击者之间的动态。该项目还提出了一种新的与攻击相关的安全性定义以及基于该定义计算信息隐藏能力的方法。最后，可以处理网络损失的鲁棒水印系统的联合源通道编码和多重描述编码方法也令人感兴趣。