摘要
对于基于音频自身特征或内容生成水印的音频水印算法,恒特征攻击是一种新的攻击方法。针对该问题,本文提出一种抗恒特征攻击的音频内容认证算法。对音频质心的计算方法做了修改,使其与密钥密切相关,提高水印系统的安全性。由修改后的质心生成水印,给出一种基于伪Zernike矩的水印嵌入方法,该方法具有较好的不可听性和容忍常规信号处理的能力。本文算法对恶意攻击能够篡改定位,同时能够抵抗恒特征攻击。与现有的音频内容认证算法相比,本文算法提高了水印系统的安全性,增大了水印的嵌入容量。
Constant-feature attack is a new method of attack for the content feature-based audio watermarking. To solve the problem, the audio content authentication algorithm robust against constant-feature attack was proposed in this paper. In the proposed algorithm, the audio centroid was modified and was made closely relat- ed to the key, which improved the security of the watermarking system. The watermark was generated by the modified centroid, and then the new watermark embedding method based on pseudo-Zernike moments was given, which was inaudible and had good ability of tolerance against some common signal processing operations. The proposed scheme not only has excellent ability of tamper detection, but also is robust against constant-feature attack. Compared with the existing audio content authentication algorithms, the algorithm improves the security of the watermarking system and increases the watermarking embedding capacity.
出处
《铁道学报》
EI
CAS
CSCD
北大核心
2013年第3期76-82,共7页
Journal of the China Railway Society
基金
国家自然科学基金(61170226
60970122)
中央高校基本科研业务费专项资金(SWJTU11CX047)
关键词
内容认证
篡改定位
恒特征攻击
质心
伪ZERNIKE矩
content authentication
tamper localization
constant-feature attack
centroid
pseudo-Zernike tooments