Improvement and Performance Evaluation of Zero-watermarking for Digital Audio Signal

• Main Authors: Hung-Shiun LIN, 林宏動
• Other Authors: Mu-Liang Wang
• Format: Others
• Language: zh-TW
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/81200241706774728779

碩士 === 樹德科技大學 === 資訊工程系碩士班 === 100 === Owning to the rapid advances of communication and information technology, lots of digital information is easily transmitted and downloaded on the Internet without getting permitted from owners. The problem of piracy is growing rapidly in recent ten years. Digital watermark embedded in digital audio can be used to verify copyright owners of the digital content, when dispute about intellectual property is happen. How to protect intellectual property is extremely important for digital content creation and development of related industry. In recent years, there are many watermarking methods have been proposed for multimedia content. Most of these methods are designed for image signals. There are relatively few in the research of watermarking of digital audio signals. As consumers demand high quality of audio, compared with traditional watermark skills, zero-watermark skills do not cause damage to the quality of digital contents. Consumers only get the private keys for digital contents, and they have rights to get the watermark to prove intellectual property. Therefore, the research of audio zero-watermark is highly valuable. In this paper, we intended to development the zero-watermark schemes for audio signal. Both “Discrete Cosine Transform (DCT) audio zero-watermarking” and “Modified Discrete Cosine Transform (MDCT) audio zeros-watermarking” are proposed in this thesis. The performance of proposed methods and the schemes present in the reference papers are evaluated. The Variety of audio signals has been used in robustness testing for these methods. The "Normalized Cross-correlation" has been measured for performance evaluation. The accuracy of reconstructed watermark of different audio zero-watermarking method is evaluated in the condition of various audio types, sampling frequency and attacks. Experimental results demonstrate that the proposed MDCT audio zeros-watermarking methods present in this thesis have excellent robustness, under the condition of the various types of signals and signal processing attacks. The purposes of protecting intellectual property and maintaining signal quality can be achieved simultaneously by using proposed methods.