改进的DCT域音频水印算法  

Improved DCT domain audio watermarking algorithm

作　　者：凡超[1] 王忠[1] 肖留威[1] 肖海[1] 

机构地区：[1]四川大学电气信息学院,四川成都610065

出　　处：《计算机工程与设计》2011年第4期1351-1355,共5页Computer Engineering and Design

基　　金：国家自然科学基金项目(49901013);中国博士后基金项目(2005038615)

摘　　要：水印的稳健性、透明性是影响水印鲁棒性的重要指标。大多数的DCT(discrete cosine transform)域音频水印算法是通过量化系数的方法嵌入水印,但是算法选取的嵌入策略、嵌入强度和嵌入位置不能达到一个满意的效果。针对这些不足,提出一种改进的DCT域量化算法。采用数据分段后隔段嵌入的嵌入策略,可以明显地改善水印的透明性;根据所分段的段能量和,采取分等级的方式确定嵌入强度,这样减少了计算复杂度;嵌入时选取高、中、低频3个系数重复嵌入;最后采用三次样条插值方法进行攻击恢复数据量。对该算法进行了Matlab仿真,通过主观和客观标准的评价,实验结果表明了该算法具有更好的稳健性和透明性。The watermarking robustness and transparency are important indicators affecting the robustness.Most of the DCT domain watermarking algorithm embed watermark by the way of quantization coefficient,at the respect of watermark robustness,transparency and computational complexity,however,the most algorithm selected embedding strategy,embedding strength and embedding position can not obtain a satisfactory result.According to these deficiencies,an improved DCT-domain quantization algorithm is proposed.Firstly,the embedding strategy of data segment,and then embedded partition can significantly improve the transparency of the watermark.To subparagraphs total energy,the algorithm determine the embedding strength by the way of energy graded reduced the computational complexity.When embedding,the paper selected the three coefficients： high,medium and low frequency,repeated embedding.Finally using cubic Spline interpolation method to restore the attacked data.After the Matlab simulation,through subjective and objective evalua-tion criteria,the simulation results show that the improved algorithm has better robustness and transparency.

关 键 词：音频水印 离散余弦变换 量化 嵌入强度 攻击恢复 

分 类 号：TN912.3[电子电信—通信与信息系统]