基于局部四元数极谐变换的鲁棒彩色图像水印方法研究

Desynchronization attacks that cause displacement between embedding and detection are usually difficult for watermark to survive. It is a challenging work to design a robust image watermarking scheme against desynchronization attacks, especially for color images. In recent years, there has been increasing research interest in robust image watermarking, and some robust image watermarking approaches that counterattack desynchronization attacks have been developed. These approaches can be roughly divided into Exhaustive search, Spread spectrum modulation, Invariant domain, Synchronization correction, and Feature-based algorithm. Through in-depth research and analysis, we observed that the former four image watermarking approaches are usually vulnerable to some desynchronization attacks, and the feature-based watermarking methods exhibit more promising than others in terms of robustness. However, there are some main drawbacks indwelled in current feature-based image watermarking schemes: Firstly, the state-of-the-art intensity based feature point detectors are sensitive to texture region, and the detected feature points are often distributed unevenly and usually not robust with respect to color variations. Secondly, the extracted local feature regions aren't fully affine invariant, and they always lose their distinctiveness when attacked by some common image processing operations. Thirdly, the nonoverlapping feature region set is not selected according to local image content. Fourthly, the existing color image watermarking schemes were designed mainly to mark the image luminance component only, and they are often sensitive to color attacks because of ignoring the correlation between different color channels. Finally, the watermark is not embedded adaptively by considering fully the color image features such as the color, edges, and region activities. This project will study the theory and method of robust color image watermarking, identify the key factors for the robustness of color image watermarking, and build the local quaternion polar harmonic transforms based overall color image watermarking system resilient to desynchronization attacks. The main contents include: Construct the stable and uniform color image feature points detector, in which the probability density and color invariance model are used; Build the fully affine invariant local feature regions based on cosine modulated Gaussian matrix and fuzzy Arimoto distribution entropy; Give the derivation of fast and accurate calculation method of quaternion polar harmonic transforms; Design the visual masking based adaptive color image watermark embedding scheme, and corresponding parallel watermarking algorithm; Study the intelligent watermark agent technology and corresponding color image watermarking software. The project will eventually build the local quaternion polar harmonic transforms based color image watermarking model, and develop the corresponding watermarking software agents.

作为图像作品版权保护的有效手段，数字水印技术已成为国际学术界研究的一个热点。但目前彩色图像水印算法异常匮乏，仅有方案只能对抗部分常规信号处理和简单全局仿射变换，尚无法有效抵抗一般性去同步攻击，即抗去同步攻击的强鲁棒彩色图像水印方法研究仍然是一项富有挑战性的工作。本项目将采用局部四元数极谐变换技术，研究高度鲁棒的彩色图像水印理论与方法，确定影响彩色图像水印鲁棒性的关键因素，建立可有效抵抗去同步攻击的强鲁棒彩色图像水印整体系统。包括构造基于颜色不变量与概率密度的高性能彩色图像特征点检测器；建立基于余弦调制高斯矩阵与模糊Arimoto分布熵的局部特征区域；推导快速高精度四元数极谐变换方法；设计基于视觉掩蔽的自适应彩色图像水印方案及其并行化算法；研究智能水印代理和基于此技术的软件系统。本项目最终将建立基于局部四元数极谐变换的抗去同步攻击彩色图像水印模型，并研制基于该模型的图像水印Agent软件包。

抗去同步攻击图像水印（特别是彩色图像水印）算法研究是一项富有挑战性的工作。本项目以“选题不大，贡献不小”为宗旨，重点关注了彩色图像水印领域的抗去同步攻击问题。本项目结合四元数变换、颜色不变量、局部信息熵等理论，研究了高度鲁棒的彩色图像水印理论与方法，确定了影响彩色图像水印鲁棒性的关键因素，建立了基于局部四元数极谐变换的可有效抵抗去同步攻击的自适应彩色图像水印整体系统，设计了可取得鲁棒性与不可感知性良好平衡的统计模型图像水印方案，初步解决了现有数字图像水印领域存在的无法有效抵抗去同步攻击的问题。.本项目属于信息安全、多媒体数据防伪领域。本项目研究成果可广泛应用于电子商务/电子政务系统、数字作品的服务付费、印刷品（包括证件、票据等）防伪、多媒体数据的分级访问、多媒体数据的跟踪和检测等诸多领域。.本项目已取得一系列高水平学术研究成果，包括：（1）获国家发明专利12项。其中，授权4项；在审8项；（2）在《ACM T MULTIM COMPUT》、《INFORM SCIENCES》、《NEUROCOMPUTING》、《J VIS COMMUN IMAGE R》、《MULTIMED TOOLS APPL》、《计算机学报》等国内外权威期刊发表学术论文29篇（其中，有27篇被SCI收录、2篇被EI收录）；（3）获辽宁省科技进步三等奖（2015）、大连市技术发明二等奖（2017）、辽宁省自然科学学术成果三等奖（2017）等多项；（4）所指导的研究生中，多人获得“辽宁省优秀学位论文”、“辽宁省优秀学位论文提名论文”称号；（5）本人于2018年10月被中共辽宁省委、辽宁省人民政府评选为“辽宁省第七批优秀专家”。本项目研究成果，有效解决了现有相关技术研究领域所存在的一系列疑难问题，进一步丰富和完善了国内外相关科学技术领域的新内容，并为相关科学技术的广泛应用提供了新途径。目前，相关研究成果已陆续被多家单位所采纳，并初步应用于电子商务/电子政务系统、数字作品的服务付费、印刷品防伪、多媒体数据的跟踪和检测等诸多领域。