摘要
提出一种图像自嵌入的水印算法 .它将一个图像的主要内容嵌入该图像自身之中 .该算法不仅能检测和定位对图像的窜改 ,而且能恢复被损坏的图像内容 .算法的基本思路是把一个图像块的主要DCT系数经过量化、编码和加密后 ,嵌入另一个图像块的最低位之中 .依据大量图像DCT量化系数的统计性质 ,本文仔细设计了主要DCT系数的编码表 ,使它既能满足水印负荷的要求 ,又能保证图像恢复时有较好的质量 .分析了水印嵌入块和原图像块间偏移值应该满足的条件和设计方法 .该算法为单向、不可逆的 。
A scheme for self-embedding an image into itself is proposed as a means for detecting and locating tamper, and self-recovering. The image is divided into small 8×8 blocks and each block is DCT transformed. A specified number of the lower frequency DCT coefficients are quantized and encoded using a fixed number of bits. The number of coefficients and their bit lengths are carefully chosen, according to the statistical property of these coefficients, the resulting bit-string for each block satisfies the payload of each block and has good self-recovering capability. The information about one block B (the bit string) is inserted into the LSB of another distant block, which is separated from B with an offset. We analyzed the method for choosing the offset. The proposed scheme is an one way watermarking and secure algorithm, and can well resist possible forged attack.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2004年第2期196-199,共4页
Acta Electronica Sinica
基金
国家自然科学基金 (No 60 0 750 0 2 )
北京市自然科学基金 (No 4992 0 0 2 )
863计划 (No 2 0 0 1AA1 4 4 0 80 )
北京市教委科技发展计划
关键词
多媒体认证
数字水印
RSA公钥算法
自嵌入
Algorithms
Cosine transforms
Image reconstruction
Image segmentation
Public key cryptography
Statistics
