一种脆弱图像水印的安全性分析与改进

Security analysis and improvement on a fragile image watermarking

分析了一种用于图像篡改检测和恢复的脆弱数字水印算法的安全性。该算法将图像划分为互不重叠的分块,通过均值比较及校验,产生各分块的认证水印,并嵌入到各自分块中。同时,在各分块中嵌入另一个分块的恢复信息,使得算法既能抵抗矢量量化攻击,又可以在某些分块被篡改时,予以一定程度的恢复。指出由于该算法生成和嵌入水印的方式过于简单,且其嵌入水印的密钥空间过小,能够通过穷举搜索轻易地发现所使用的密钥。因而能够伪造水印,导致错误认证。通过对于嵌入的水印进行混沌调制和置乱,并扩展密钥空间,有效地改进了算法的安全性。

The security of a fragile digital watermarking scheme for image tamper detection and recovery is analyzed. The watermarking scheme divides an image into non-overlapping blocks, generates the authentication watermark for the blocks with comparison and parity check between the average intensities and embeds them into their own blocks. Also, the recovery information of another block is embedded into each block, which makes the scheme can resist an vector quantization （VQ） attack, while recover the image to a certain degree when some blocks are tampered. It is indicated that since the method for watermark generation and embedding is too simple, while the key space for watermark embedding is too small, the watermark embedding key can he obtained easily with exhaustive searches. Therefore, one may counterfeit watermark and lead incorrect authentication. The security of the scheme is effectively improved by using chaotic modulation and permutation to the em bedded watermark and making a expansion for the key space.