【摘要】 不可见性和鲁棒性是用于版权保护的数字水印算法的研究重点。本文结合人类视觉系统特性,利用自组织映射网络对载体图像子块进行特性分类,然后将加密水印DCT域奇异值自适应地嵌入到载体图像子块的DCT域的中频系数中。实验证明,由于自组织映射网络良好的分类特性和SVD的特性,该算法具有良好的不可见性和鲁棒性。更多还原

【Abstract】 Invisible and robust characteristics are emphases of digital watermarking algorithm for copyright protection.First,we use SOM neural network to classify the blocks of the original image based on the HVS.Second,we encrypt the gray watermark and apply DCT to it,then apply singular value decomposition.Third,Watermark’s singular values with different strength are inserted into the Mid-frequency coefficients of DCT domain according to the classification.Due to the better characteristics of SOM neural network classification and SVD,our algorithm has better invisible and robust characteristics.It is proved by the experimental results.更多还原