Current Issue Cover


摘 要
目的 基于数字水印技术的音乐作品版权保护是目前国际学术界的研究热点之一。但目前绝大多数数字音频水印方案仅仅能够对抗简单的常规信号处理,尚无法有效抵抗破坏性较强的一般性去同步攻击,即抗去同步攻击的强鲁棒数字音频水印理论与方法研究仍然是本领域的研究热点。方法 提出了一种基于稳健局部特征的非下采样小波域数字水印算法。该算法首先利用非下采样小波域平滑梯度检测算子从载体音频中提取出稳定的音频特征点;然后结合数字音频样本响应确定出局部特征音频段;最后采用量化调制策略将数字水印信号重复嵌入到局部特征音频段中。结果 本文选用四段典型的采样频率为44.1kHz、量化精度为16比特、长度为15秒的单声道数字音频信号作为原始载体进行测试,并与经典算法对比了不可感知性和鲁棒性。本文算法在含水印音频与原始载体音频间的信噪比上平均提升了5.7dB,同时常规攻击和去同步攻击下的平均检测率分别保持在0.925和0.913,高于大多数传统算法,证明了本文算法具有较好的不可感知性,并且对常规信号处理(MP3压缩、重新量化、重新采样等)和去同步攻击(包括幅度缩放、随机剪切、音调伸缩、DA/AD转换、抖动等)均具有较好的鲁棒性。结论 本文利用描述能力强且性能稳定的平滑梯度刻画局部数字音频性质,进而提出一种基于平滑梯度的非下采样小波域音频特征点提取方法,有效解决音频特征点稳定性差且分布极不均匀的缺点,提高了数字音频水印对音调伸缩、随机剪切、抖动等攻击的抵抗能力。
A digital watermarking algorithm in UDWT domain based on robust local feature

niupanpan,yangsiyu,shen xin,yanghongying,shiqiliang,wangxiangyang()

Objective Facing the ever-growing quantity of digital documents transmitted over the internet, it is more than ever necessary for efficient and practical data hiding techniques to be designed in order to protect intellectual property rights. Digital watermarking techniques have historically been used to ensure security in terms of ownership protection and tamper proofing for a wide variety of data formats. This includes images, audio, video, natural language processing software, relational databases, and more, this paper focuses on audio watermarking. Generally, digital audio watermarking is the technology of embedding a useful data (watermark data) within a host audio and the perceptual quality of the host audio should not be degraded substantially by the embedding. For different purposes, audio watermarking can be branched into two classifications: robust audio watermarking and fragile audio watermarking. Robust audio watermarking is used to protect ownership of the digital audio. In contrast, the purpose of fragile audio watermarking is digital audio authentication, that is, to ensure the integrity of the digital audio. For a digital watermarking scheme, there are generally three major properties, namely imperceptibility, robustness, and payload. Imperceptibility means that the watermarked audio is perceptually indistinguishable from the original one. This is needed to maintain the commercial value of audio data or the secrecy of the embedded data. Robustness refers to the ability that the watermark survives various attacks, like JPEG/MP3 compression, additive noise, filtering, and amplifying etc. Payload refers to the total amount of information that can be hidden within the digital audio. Imperceptibility, robustness, and payload are three main requirements of any digital audio watermarking systems to guarantee desired functionalities, but there is a tradeoff among them from the information-theoretic perspective. Improving the ability of robustness, imperceptibility, and payload at the same time has been a challenge for a digital audio watermarking algorithm. A digital audio watermarking scheme must be robust against a variety of possible attacks. Attacks which attempt to destroy or invalidate watermarks can be classified into two types, noise-like common signal processing operations and desynchronization attacks. Desynchronization attacks are more difficult to tackle than other types of attacks. It is a challenging work to design a robust digital audio watermarking algorithm against desynchronization attacks. Method In this paper, we proposed a new second generation digital audio watermarking in undecimated discrete wavelet transform (UDWT) domain based on robust local audio feature. Firstly, robust audio feature points are detected by utilizing smooth gradient, and these feature points are always invariant to common signal processing operations and desynchronization attacks. Then, local digital audio segments, centering at the detected audio feature points, are extracted for watermarking use. Finally, the watermark is embedded into local digital audio segments in UDWT domain by modulating the low-frequency coefficients. We employ the robust significant UDWT coefficients, which can effectively capture the important audio texture features, for locating accurately the watermark embedding/ extraction position, even suffering the desynchronization attacks. Results In order to evaluate the high performance of our scheme, watermark imperceptibility and robustness test are illustrated for the proposed watermarking algorithm, and the proposed watermark detection results are compared with some state-of-the-art audio watermarking schemes against various attacks under an equal condition. All of the audio signals in the test are music with 16 bits/sample, 44.1 kHz sample rates, and 15 seconds. All our experiments are accomplished on a personal computer with Intel Core i7-4790 CPU 3.60GHz, 16GB memory, and Microsoft Windows 7 Ultimate operating system. Also, MATLAB R2016a is used to perform the simulation experiments. To quantitatively evaluate the imperceptibility performance of the proposed watermarking algorithm, we also calculate the SNR, which is an objective criterion and is always used to evaluate audio quality. The SNR of the proposed scheme is improved by 5.7dB on average respectively, which shows the effectiveness of the proposed scheme in terms of the invisibility of the watermark. Watermark robustness is measured as the correctly extracted percentage of extracted segments. And the average detection rate remained at 0.925 and 0.913, respectively, higher than most traditional algorithms. Therefore, the experimental results show that the proposed approach has not only good transparency, but also has strong robustness against common audio processing such as MP3 compression, resampling, re-quantization, and good robustness against the desynchronization attacks such as random cropping, pitch-scale modification, jittering et al. Conclusions Audio watermarking algorithm based on robust feature point of wavelet domain. This algorithm is proposed based on content feature of audio and stability of low frequency coefficient of UDWT. First, original audio is dealt with by way of UDWT, then calculating first-order gradient responses of low frequency coefficient, ranking those responses in a descending order and choosing the highest response as criteria to set threshold. From this, stable and evenly distributed feature point is achieved and then setting the robust feature point as identification, inserting audio watermarking. In the end, low frequency coefficient is inserted into watermarking by way of QIM. The proposed scheme effectively solves the drawbacks of poor stability and uneven distribution of audio feature points, and improves the resistance of digital audio watermarks to pitch-scale modification, random cropping, and jittering attacks.