医学图像的高容量及鲁棒可逆水印的研究

Reversible watermarking works effectively for medical images content authentication by embedding a watermark (such as identification data, serials number, text or image etc.) to medical images without influencing the vision of medical images in network environment. And meanwhile, reversible watermarking can reconstruct the original image without any distortion after the hidden data have been extracted; thus, reversible watermarking is very useful, especially in applications dictating high fidelity of multimedia content, such as medical images. In recent years, higher and higher requirements are raised for embedding-capacity and robustness when reversible watermarking is used for medical images. In combination with self-characteristics of medical images, this project is to study four aspects below: ① the theoretical limits of the capacity are investigated under the condition that the embedding distortions are controlled in a small range when reversible watermarking is used for medical images. ② all the regions being capable of carrying high capacity are obtained, and then the self-adaptive embedding strategy is proposed to self-adaptively embedding as much watermarking information as possible into these regions with high capacity. ③ the key idea of the pre-processing method is to largely increasing the embedding capacity by correspondingly decreasing the embedding distortions introduced by reversible watermarking, in this project, the self-adaptive pre-processing strategy is proposed to self-adaptively select the appropriate pre-processing algorithm to reduce the embedding distortions introduced by reversible watermarking. ④ research on improving the robustness of reversible watermarking used in medical images, the bijective integer transform (or an statistical invariant and its corresponding embedding method) resistant to the attacks is proposed. By study on the above problems, disadvantage will be improved when reversible watermarking is used for the medical images. The great contribute will be done to the theoretical and algorithmic research for reversible watermarking of medical images. Research results will also further enhance the practical application of medical images.

可逆水印技术为网络环境下的医学图像的版权保护提供一种有效的保护手段。且可逆水印技术能很好满足医学图像不允许失真的特殊要求。近年来，为了解决医学图像的应用局限性，对医学图像的可逆水印在容量和鲁棒性上提出了越来越高的要求。结合医学图像的自身特点，本研究具体包括四个方面 ① 理论研究可逆水印用于医学图像时所能达到的容量上限；② 确定同等失真下携带载荷能力强的区域，提出基于图像统计特性的自适应嵌入策略；③ 以预处理的方式降低嵌入失真来换取容量的提高，提出新的自适应图像的预处理方法；④提出对某些攻击具有不变性的双射的整数不变域（或空域统计不变量及其嵌入方式），提高医学图像的可逆水印的鲁棒性。本项目的研究成果将进一步推动医学图像的可逆水印的理论和算法研究，极大促进可逆水印在医学图像中的实际应用。

摘要：本项目以医学图像远程医疗为应用背景，研究可逆水印算法性能提升和抵抗攻击的能力。具体内容包括: ①研究如何在保持载体图像视觉质量的情况下显著提高水印嵌入的数据容量，即研究高容量可逆水印算法，建立空域失真模型，提取最能描述局部复杂度的特征，按照复杂度的强度进行自适应水印嵌入；②研究在给定嵌入容量（小容量）的情况下如何更有效地提升图像视觉质量：研究基于局部相关性的精确定位准则，研究基于图像内容的自适应预测器，研究自适应嵌入策略和双层嵌入技术，研究能获得最佳嵌入区域和熵更小（分布更集中）的预测误差直方图的可逆水印方案；③ 提出了基于角度差的量化索引调制（difference angle quantization index modulation，DAQIM）方法，该方法的目的是提高抵抗增益攻击的能力，实验结果证明，我们提出的方法在对各种攻击的鲁棒性的方面优于常用的现有方法，例如JPEG量化噪声，加性高斯白噪声(AWGN)，裁剪效果和均值滤波等攻击。.发表了学术论文16篇，其中SCI期刊论文5篇，EI期刊论文4篇；录用SCI期刊论文4篇，二审小改SCI期刊论文1篇，一审小改SCI期刊论文1篇，投稿SCI期刊论文2篇；获发明专利1项,申请发明专利4项。