一种改进的基于奇异值分解的数字水印算法

Improved SVD-Based Digital Watermarking Algorithm

提出了一种基于奇异值分解(SVD)的数字图像水印改进算法。该算法将原始图像进行分块,对前M个子块进行奇异值分解,之后选取各子块中最大的奇异值,通过量化的方法嵌入经过BCH纠错码及交织预处理后的M个水印信息。由于采用了量化的方法嵌入水印信息,其提取不需要原始图像的任何信息,实现了水印的盲提取。同时,提取过程受到交织参数的限制,不知道该参数就无法正确地提取出数字水印,实现了水印的加密。实验结果表明,该改进算法明显改善了原算法的鲁棒性和安全性。

An improved SVD-based digital watermarking algorithm was proposed, which made the original image divide into blocks and brought the former M sub-blocks into singular value decomposition. Then each sub-block＇s biggest singular value was selected. After BCH error correcting codes and interleaved pretreatment the M watermark information was embedded into these biggest singular values by quantization method. Because the watermark information embedded was done with quantization methods, the received image authentication needed no information about the original image. This algorithm realized blind extraction of digital watermarking. Meanwhile, the extraction process was effected and limited by the interleaving parameter, so the digital watermarking could not be extracted without this parameter. The watermark was encrypted. The experimental results showed that the improved algorithm achieved better robustness and safety than that of the original one.