An Audio Aggregation Watermarking For Copyright Protection And Integrity Authentication

With the rapid development of computer communication technology and the internet, the spread of digital media has become more and more conveniently, which makes digital media be easy to be copied and pirated. This circumstance damages the rights of works'owners and impacts the normal service market of multimedia, therefore, it is urgent to protect the copyright of the digital media. In recent years, digital watermarking technology which is used for protecting copyright of the digital media effectively has been widely researched and applied. Digital watermarking technology is also applied for integrity Authentication and the other area. The aimed carrier of watermarking technology contains image, audio, video, text and other multimedia works. In generally, the existed audio watermarking technology is just used for a single audio, however, audio works are often published in the form of audio album(an aggregation), in view of practical needs, in this paper, we will probe and research the watermarking technology which is applied for audio aggregation.The watermarking technology in this paper based on audio aggregation contains some algorithms and features as follows:1, Digital audio watermarking technology, basic theory, and development status was systematically introduced, according to analyze existed audio watermarking algorithm and watermarking framework, we presented audio aggregation watermarking technology from a new point, introduced the concept of audio aggregation watermarking and algorithm model. And in terms of audio aggregation's particularities, we probe some attacks aimed at audio aggregation.2, According to the fuzziness and randomness of cloud model, we proposed an audio aggregation algorithm based on Cloud model. Wavelet coefficients of original audio aggregate are mapped into Cloud drops, Cloud drops is zero-watermark. And then make use of Reverse-Cloud generator to calculate three numerical characteristics of Cloud model, these characteristics are used for assessing similarity of original watermark and existed watermark. After the attacks, a part of cloud drops have been changed, which has little impact on characteristics of the entire cloud model, so utilize Change rate of these characteristics to detect watermark, the robustness is strong. 3, According to lift-wavelet, we propose an algorithm which not only protects copyright of a single audio but also protects copyright of the whole audio aggregation. Using fast lifting wavelet decomposition technique to extract features from every audio work in aggregation, the features are constructed as zero-watermark of every audio, which is used to implement copyright protection of corresponding audio; compose all zero-watermarks of all audio works into a zero-watermark matrix to get aggregate zero-watermark, which is used to implement copyright protection of the audio aggregation.4, Propose an audio watermarking algorithm based on ICA for implementing copyright protection of audio aggregation, the wavelet coefficients of each audio work in the audio aggregation are set as observation data, and independent components of the observation data are extracted using ICA separate matrix, and then two most important components are chosen to embed the common watermark; ICA is used to detect watermark, it is not necessary to know the embedding location and strength of watermark and other factors5, Propose an audio watermarking algorithm based on ICA for implementing integrity authentication, audio aggregation are put into wavelet packets decomposition, coefficients are extracted by ICA technology, and then a right component are chosen to be embedded watermark. The algorithm not only implements integrity authentication for audio aggregation but also recovers attacked audio aggregation.6, Improve the (k, n) method proposed by Shamir firstly, and propose an audio aggregation watermarking algorithm based on (k, n/k=n) for implementing integrity authentication. Every pixel in watermark image are distributed by (k, n) method, and then embedded into every audio work in audio aggregation, so one watermark image are embedded into an audio aggregation, when extracting watermark, set k=n, once there are audio works in aggregation are attacked, the extracted watermark image will change correspondingly, which enhances fragility of the algorithm.