【摘要】 提出了一种针对MPEG-2 AAC压缩音频的脆弱水印算法.该算法利用了AAC中MDCT量化系数大于15的Huffman编码特性来嵌入水印,并依据水印的频率分布和音频信号感知熵PE的大小对嵌入算法进行了分析和改进.实验表明,该算法具有较高隐藏率和良好的不可感知性,并且水印的嵌入和提取过程十分方便快速,适合进行实时分析.更多还原

【Abstract】 A fragile audio watermarking method for MPEG-2 AAC bitstreams was proposed.The algorithm of watermark embedding was carried out by modifying the LSBs(least significant bits) of quantized MDCT coefficient’s magnitude that is greater than 15,and was improved by considering the watermark’s distribution in frequency and the perceptual entropy of host audio.Experimental result shows the proposed watermarking scheme is not appreciable,and it is possible to insert high hide rate additive data effectively into encoded bitstreams.The embedding and retrieving process is easy and it is adaptive for real-time analysis applications.更多还原