数字水印提取的迭代学习辨识方法  

Iterative Learning Identication for Digital Watermark Extraction

作　　者：邹凡[1] 孙明轩[1] 毕宏博[1] 

机构地区：[1]浙江工业大学信息工程学院,杭州310023

出　　处：《科技通报》2011年第5期691-695,699,共6页Bulletin of Science and Technology

基　　金：国家自然科学基金项目(60874041)

摘　　要：针对结合混沌映射和多重混合的数字图像水印嵌入及提取问题,提出利用迭代学习辨识算法提取水印图像的方案。该方案将水印掩盖于混沌系统的参数中,以迭代学习算法重建原始水印图像。实验表明,运用所提出的迭代学习辨识算法能够从含水印图像中提取出完整的水印图像,给出的测试结果验证了学习算法抵抗常见几何攻击的能力。This paper presents a digital image watermarking scheme based on chaotic mapping and multi-blending,and the decryption and extraction of the digital watermark is carried out from iterative learning identi？cation perspective. With the proposed scheme the watermark image is added to a parameter of the chaotic system undertaken,and the mechanism of multi-blending is used for generating the final image,which aims to enhance watermarking security. An iterative learn- ing identification algorithm is proposed for the recovering purpose. Experimental results demonstrate that the digital im- age watermarking can be extracted completely,and the learning algorithnl has an ability to resist various types of attacks.

关 键 词：迭代学习辨识 混沌映射 多重混合 数字水印 

分 类 号：TP309.7[自动化与计算机技术—计算机系统结构]