一個(gè)改進(jìn)的基于直流電平偏移的音頻水印算法

一個(gè)改進(jìn)的基于直流電平偏移的音頻水印算法

根據(jù)水印調(diào)整直流電平偏移的音頻水印算法具有很好的低通和杭噪性能當(dāng)水印發(fā)生去同步時(shí)，如果同步點(diǎn)偏移不大，直流電平的符號(hào)不會(huì)改變，但當(dāng)同步點(diǎn)偏移很大時(shí)，直流電平可能變號(hào)，從而導(dǎo)致水印提取錯(cuò)誤。為此，提出了在宿主音頻中嵌入同步訓(xùn)練序列的抵制去同步攻擊的改進(jìn)算法實(shí)驗(yàn)結(jié)果表明，算法對(duì)隨機(jī)裁剪等去同步攻擊的魯棒性得到了明顯提高.

A udiow atermarkingt hatm odulatesD C levels hift accordingt o watermark has wonderful low pass and noise resistance performance . When loss of watermarking synchronization occurs, if the shift of synchronization position is trivial, the sign of DC level will not be changed, but if notable, the sign of DC level will be changed, as result watermarking extracting error occurs. In order to resist de-synchronization attacks, an improved algorithm that embeds training sequence in cover audio is proposed. Experimental results show that this technique obviously improves the watermarking robustness resistance to de-synchronization attacks.