数字音频信号的脆弱水印嵌入算法

该文给出了量化系数嵌入水印比特信息的基本原理，在此基础上提出了数字音频信号的脆弱水印嵌入算法。该算法充分利用离散子波变换的多分辨率特性，通过等概率随机量化音频信号不同子带的子波系数嵌入视觉可辨别的水印－二值图像。仿真实验结果表明，该文提出的数字音频信号脆弱水印嵌入算法对有损压缩、滤波和重新采样等操作具有很强的敏感性。

The Fragile Watermark Embedding Algorithm for Digital Audio Signal

With the development of computer, network and multimedia techniques, digital products of text, image, audio, and video have been popularized increasingly and applied widely. However, the issues of copyright protection and information security have become more serious than ever due to the characteristics of digital multimedia, such as case of coping, editing, storing and transmitting through networks. In general, digital watermarking audio signal has been used to protect its copyright or verify its originality. The embedding algorithm of digital audio signal can be classified into two groups: robust algorithms and fragile algorithms. In the later group, the embedded watermark must be inaudible and sensitive to several operations performed on audio signal. This paper presents a novel fragile watermark embedding algorithm of digital audio signal. The algorithm is based on the principle of embedding watermark bit information by quantizing coefficient. The embedded watermark is a visually recognizable image. Firstly, digital audio signal is multi resolution decomposed by wavelet transform. Secondly, the two dimensional watermark is decreased to a one dimensional sequence. Thirdly, the wavelet coefficients are randomly selected from the different wavelet decomposition levels and quantized for embedding all elements of the watermark sequence. Finally, the watermarked audio signal is recovered by inverse wavelet transform. The proposed algorithm in this paper makes good use of the multi resolution characteristics of wavelet transform. Experimental results show that the algorithm is inaudible and sensitive to several operations, such as lossy compression, filtering, re sampling, performed on audio signal.