基于多描述编码的自恢复脆弱水印算法

摘要

兼顾水印嵌入容量和篡改恢复质量，基于多描述编码提出一种自适应图像内容的可恢复脆弱水印算法。该算法在图像块自适应分类的基础上，将图像块的DCT系数分为两个独立且互补的变长描述编码，每个描述能独立地近似恢复图像块的内容，两个描述编码一起得到的恢复质量更高。水印嵌入时，两个描述编码基于密钥随机嵌在不同图像块中，以提高算法在不同攻击下的篡改检测性能。被篡改图像块，只要有一个描述有效即可恢复该篡改块，若两个描述同时有效，可进一步提高篡改块的恢复质量，有效缓解了同步攻击与数据浪费之间的矛盾。实验结果证明，该算法不仅有效降低了生成的恢复水印容量，同时提高了不同攻击下的篡改检测性能和篡改恢复质量。%To take watermark capacity and restoration quality into account,a kind of recoverable fragile water-marking scheme for adaptive image contents based on multiple description coding was proposed.For each block,two description codes with different lengths were generated according to its DCT coefficients based on a-daptive classification of image blocks.Each description code could independently and approximately restore the image block content and two description codes could achieve higher quality of the recovered block.In the water-mark embedding stage,two description codes were randomly embedded into other image blocks based on the secret key to improve the tamper detection performance under different attacks.The tampered image block could be approximately restored as long as there was a valid description code.If there were two valid descrip-tion codes of the tampered blocks,the quality of the recovered block of it could be further improved and the conflict of tampering coincidence and watermark-data waste could be effectively relieved.Results demonstrate that the proposed method effectively reduces the watermark capacity,and improves the tamper detection per-formance and the recovery quality under different attacks.