图像可逆变换机制及关键技术研究

Recently, the topic of reversible data hiding has received intensive study. However, the applicability of reversible data hiding in the real world has not been seriously taken into consideration. In this project, the research work conducted home and abroad on image reversible transformation will be investigated. Based on the techniques of reversible data hiding, the mechanism of image reversible transformation will be studied so that the meaningful transformation techniques in the specific scenarios will be developed. More specifically, we will focus on reversible contrast enhancement, reversible visual transformation, reversible transformations for the compressed images, and image reversible transformations in the homomorphic encryption domain. According to the related theories, we will integrate a number of key techniques involved in image reversible transformation, such as reversible data hiding, image quality assessment, visual information processing, secret message transmission, privacy protection and image enhancement, trying to achieve the reversibility while obtaining the desired effects of image transformation in the real applications. Based on the work that has been accomplished, we will construct the theoretical model of image reversible transformation, trying to find out the prerequisite of performing the reversible transformation and the upper bound of algorithm performance. Moreover, the theoretical analysis will help to guide the algorithm design and performance optimization. Through the study of this project, the virtue and technique level of image reversible transformation will be raised, while its applicability will be broadened. Furthermore, the advances made in this project on image reversible transformation will set up examples and references for the other research topics in information hiding and cyberspace security.

近年来，可逆数据隐藏技术得到了深入研究，但其实用性却未受到充分关注。本项目在国内外研究成果的基础上，基于可逆数据隐藏的技术手段，研究图像的可逆变换机制，并根据具体的应用场景开发有意义的图像可逆变换技术。具体而言，重点研究图像的可逆对比度增强、可逆视觉变换、压缩图像的可逆变换以及同态加密域中的图像可逆变换等多项关键技术问题。结合相关技术理论，整合与图像可逆变换有关的可逆数据隐藏、图像质量评估、视觉信息处理、秘密消息传送、隐私保护、图像增强等诸多课题，在保证变换可逆性的同时取得实际应用中所需的变换效果。此外，将在现有工作的基础上创建图像可逆变换空间的理论模型，探讨图像可逆变换能够实施的客观条件和性能上限，用于指导算法设计和性能优化。通过本项目的研究，将显著提高图像可逆变换的实用价值和技术水平，拓宽其应用范围，取得的进展将会为信息隐藏乃至网络空间安全中的其它课题提供有益的参考和借鉴。

本项目围绕数字图像（包括自然图像和医学图像）可逆对比度增强，加密域（包括同态加密和流加密系统）可逆数据隐藏，可逆图像视觉变换（包括空域图像和压缩图像）等课题开展研究。项目期间提出了多种图像可逆变换和加密域可逆数据隐藏算法，拓展了可逆数据隐藏的应用场景，取得了一系列成果。课题组利用项目资助发表期刊论文15篇（其中14篇被SCI收录，1篇被EI收录），会议论文4篇（均被EI收录），申请专利3项，培养硕士研究生8名，协助培养博士研究生1名。项目研究成果具有一定的科学意义，得到了同行和学术界的关注，据科研之友统计发表的论文被引用81次，取得了一些独创性成果和进展。