To explore the principle and performances of Quantization Index Modulation (QIM) watermarking algorithm, the realization schemes of this algorithm in Discrete Fourier Transform (DFT) domain are classified into three types according to the differences of quantizer parameters, and then the illustrative diagrams of these schemes are given, so the properties of transparency, robustness and capacity of each scheme are concluded. The scheme with best distortion-robustness trade-off is found on that basis. The more efficient embedding and extracting formulas of this scheme are given, and the performances are researched by theoretical analysis rather than computer simulation. The conclusions show that the transparency is in proportion to the square of quantization step and small DFT coefficient leads to poor transparency. The robustness to Additive White Gaussian Noise (AWGN) only relates to the quantization step and the algorithm is susceptible to amplitude scale and resampling attacks. The possible capacity is F_o/2 bits per second, which can satisfy various watermarking applications. To improve the robustness, an adaptive quantization algorithm based on masking properties of the Human Auditory System (HAS) is proposed at the end of this paper, and the algorithm is evaluated in the analog channel environment and its robustness is proved by the results.

No abstract

No abstract

To analyze the robustness performance of direct sequence spread spectrum (DSSS) based audio watermarking algorithm theoretically, the fundamental principle of spread spectrum watermarking is analyzed firstly. The communication model of DSSS-based audio watermarking system is designed according to the spread spectrum communication theory. The interrelation between bit error rate (BER) and watermark power and processing gain is revealed by utilizing definition of the optimal baseband transmission system under ideal channel. Finally, Monte Carlo simulation method is used to prove the BER formula. Experiments reveal that the expression can calculate the BER of detected watermarking under cover audio and additive white Gaussian noise (AWGN) interferences accurately. Keywords-audiowatermarking; direct sequence spread spectrum; bit error rate; processing gain 基于直接序列扩频的音频水印算法鲁棒性能分析 陈光，杨森斌，张倩，杜佳 西安通信学院，西安，中国，710106 ysb-007@163.com 【摘要】为从理论上评估基于直接序列扩频的音频水印算法的鲁棒性能，分析了扩频音频水印的基本原理，结 合扩频通信理论，设计了加性高斯白噪声攻击下扩频音频水印系统的通信模型。依据理想信道下最佳基带传输 系统的定义和公式，推导了水印信息误比特率与嵌入强度、处理增益之间的关系。最后利用 Monte Carlo 仿真 方法，对公式进行了验证。实验结果表明，该公式可精确计算出载体音频和加性高斯白噪声攻击下水印检测的 误比特率。 【关键词】音频水印；直接序列扩频；误比特率；处理增益 1 引言 近年来，为保护数字音像作品的版权，数字音频水印 技术得到国际科技界和企业界的广泛重视，众多水印算法 被提出并得到应用。其中，基于直接序列扩频(Direct Sequence Spread Spectrum, DSSS)的音频水印算法 [1] (以 下简称为扩频音频水印算法)由于具有鲁棒性强、隐蔽性 好以及安全性高等优点，已成为数字水印研究领域的经典 算法和研究热点 [2-5] 。 目前，对扩频音频水印算法的研究主要集中于算法性 能改进(如利用听觉掩蔽效应改善透明性 [2] 、优选扩频码 增强鲁棒性 [3] 、 采用软扩频或 CDMA 技术增大隐藏容量 [4,5] 等)以及算法实现应用 [3,5] 两方面，而对信息隐藏模型、性 能分析等基础理论的研究相对较少。此外，由于缺乏公平 统一的性能测试与评价体系，大多数文献在研究算法性能 时都采用仿真实验手段。然而，由于使用的载体音频和攻 击方法各不相同，实验结果的可信度和可比性较差。 本文在分析扩频音频水印算法基本原理的基础上，设 计了加性高斯白噪声( Additive White Gaussian Noise, AWGN)攻击下扩频音频水印系统的通信模型，从扩频通 信理论的角度，得出了误比特率与水印嵌入强度、扩频处 理增益之间的公式，并通过蒙特卡罗实验进行了验证。利 用该公式， 可从理论上分析扩频音频水印算法的鲁棒性能， 得出加性高斯白噪声攻击下水印检测的误比特率，也可为 算法参数选取提供依据。 2 扩频音频水印算法的基本原理 978-1-4244-7255-0/11/$26.00 ©2011 IEEE

To explore digital watermarking specifically suitable for the speech domain, this paper experimentally investigates the properties of line spectrum pair (LSP) parameters firstly. The results show that the differences between contiguous LSPs are robust against common signal processing operations and small modifications of LSPs are imperceptible to the human auditory system (HAS). According to these conclusions, three contiguous LSPs of a speech frame are selected to embed a watermark bit. The middle LSP is slightly altered to modify the differences of these LSPs when embedding watermark. Correspondingly, the watermark is extracted by comparing these differences. The proposed algorithm's transparency is adjustable to meet the needs of different applications. The algorithm has good robustness against additive noise, quantization, amplitude scale and MP3 compression attacks, for the bit error rate (BER) is less than 5%. In addition, the algorithm allows a relatively low capacity, which approximates to 50 bps.