适用于JPEG图像的高保真鲁棒可逆信息隐藏算法研究

Conventional reversible data hiding (RDH) mainly focuses on the research for uncompressed images, and may undergo a drawback of limited uses in practical applications. As the most used medium, the research on the JPEG image is not paid enough attentions by the RDH community, and therefore cannot meet the high-fidelity embedding requirement as it lacks a solid theoretical analysis. This problem is particularly prominent in the contradiction between the increasing demand of secure "hidden behavior" over social media and the slow progress of corresponding research with insufficient investigations. To address these issues, we propose to tailor the RDH for JPEG images with new techniques, including the frequency band combination-based multiple histogram generation, quantization distortion model-based adaptive data embedding and the file size preservation for the marked image, and will take both the theoretical and practical evaluations into account to solve the optimization problem for a given capacity, and some effective high-fidelity algorithms will be given. Based on the exploitation of JPEG, we will formulate a novel robust RDH framework with stable performance from a new prospective in terms of both the theory and the motivation. All the proposed techniques will be encapsulate into a generalized framework. It is expected that the proposed research will provide the theoretical significance and practical value for development of RDH.

传统可逆隐藏技术研究主要集中于无压缩图像方面，但应用面窄，而针对实际应用的JPEG图像研究却比较少，相关研究尚难以实现JPEG图像的高保真信息隐藏需要，也未构建足够坚实的理论根基。这一问题在当前社交媒体逐步兴起、“行为隐藏”日益重要而相关可逆研究进展缓慢、投入不足的矛盾下显得尤为突出。本项目将针对JPEG图像的可逆信息隐藏领域中的亟待解决的问题，通过基于频带融合的多直方图生成、基于量化失真模型的自适应嵌入和含密图像体积保持技术研究，兼顾理论和实用评价标准，解决等容量下的优化问题，设计具备JPEG图像特点的多种高保真可逆算法，并以此为基础探索具备一定鲁棒性的JPEG可逆嵌入的新理论、新思路，实现JPEG图像的高保真可逆嵌入和具备稳定鲁棒性的一般性可逆框架，为推进可逆信息隐藏研究的发展和实际应用提供理论和技术参考。

本课题以最广泛应用的压缩图像为载体，研究适用于JPEG图像的高保真鲁棒可逆信息隐藏算法，兼顾理论和应用标准，推进该领域的理论创新，扩展可逆算法的应用面。课题以基于频带融合的多直方图生成、基于量化失真模型的自适应嵌入和含密图像体积保持技术研究为研究内容，四年来提出了高容量多直方图修改嵌入技术、二维多直方图自适应修改技术、人眼视觉评价优化嵌入、JPEG码流可逆优化映射技术、高保真鲁棒可逆算法框架综述，以及可逆JPEG技术融合实际应用的医学图像信息隐藏算法。从实验结果来看，所得研究成果能够有效提升JPEG图像的嵌入性能，所提理论能够进一步融合JPEG编码特性为设计自适应的嵌入算法提供支撑。此外，从综述研究看，新的算法框架展现出与先进人工智能研究有机结合的强大潜力，为进一步研究指明了方向。