摘要
先对载体图像进行分块,然后对每块二级LWT后的中高频带再次LWT,并对选取的各频带进行SVD,选取其相应的奇异值组成新的变换矩阵,对新的变换矩阵按规则进行分块,并再次SVD。通过两次分块,两次LWT和四重使用SVD构造矩阵的方法,有效地将抽取的奇异值重新分配和组合,最后将Arnold置乱后的灰度水印信息加载到组合后的矩阵中。仿真实验表明,新的图像水印方法提高了嵌入的信息量,具有很好的透明性,而且对常见的图像攻击具有很强的鲁棒性。
The original image is divided into many blocks.Then the middle-high sub-bands of two-level LWT with each block are implemented with LWT,and SVD is applied to the selected sub-band,the new matrix is constructed by singular values,which is divided into many blocks by rule,and decomposing with SVD again.Through twice partition blocks,twice LWT and the use of quartic SVD,the matrix will re-allocation and combination for extraction＇s singular value,and finally the watermarking image with Arnold scramble is embedded into the matrix.Experimental results show that the proposed method increases embedded capacity,and has a good transparence and robustness against attacks for common image processing.
出处
《计算机工程与应用》
CSCD
北大核心
2010年第11期104-106,108,共4页
Computer Engineering and Applications
基金
国家自然科学基金No.50175070~~
关键词
提升小波变换
奇异值分解
ARNOLD置乱
透明性
鲁棒性
Lifting Wavelet Transform（LWT）
Singular Value Decomposition（SVD）
Arnold scramble
transparence
robustness
