CAJ | 학술논문

提出一种新的基于DCT系数比较的音频盲水印算法。该算法依据平均值具有很好的鲁棒性特点，将音频信号按水印长度分为若干个不重叠数据段，对每个数据段取绝对值的平均值， n个数据段绝对值的平均值一组进行DCT变换，然后以DCT系数中的次小值作为参照，修改DCT系数中的最小值实现水印信息的嵌入。实验表明，该算法不仅具有较好的不可感知性，对压缩类攻击也具有很强的鲁棒性。
제출일충신적기우DCT계수비교적음빈맹수인산법。해산법의거평균치구유흔호적로봉성특점，장음빈신호안수인장도분위약간개불중첩수거단，대매개수거단취절대치적평균치， n개수거단절대치적평균치일조진행DCT변환，연후이DCT계수중적차소치작위삼조，수개DCT계수중적최소치실현수인신식적감입。실험표명，해산법불부구유교호적불가감지성，대압축류공격야구유흔강적로봉성。
In this paper we present a new blind audio watermarking algorithm,which is based on DCT coefficients comparison.Based on the characteristic of mean which has very good robustness,the algorithm divides the audio signal into several non-overlapping data segments ac-cording to the length of the watermark,and calculates the mean of absolute value for each data segment,and takes the means of absolute values of n data segments as a group to apply DCT transform.Then it modifies the minimum value of DCT coefficients by taking the second smallest value in DCT coefficient as the reference to implement the embedment of watermarking information.Experiments show that the method has good imperceptibility as well as strong robustness to the compression attacks.