以基因演算法為基礎之強健式音訊浮水印

Abstract

數字化年代的來臨，為多媒體帶來數字化的新革命，也讓多媒體信息的存取更便利、更有效和更準確。隨著Internet的普及化，多媒體作品一方面便利於交流一方面也讓惡意侵權和篡改更容易。利用科學技術，在Internet的便利環境下，利用浮水印技術，有效地保護智慧財產權，已變成一個重要的課題。本文提出一種加入版權訊息的方法，一旦發生版權糾紛，可以把裏面的隱藏訊息解出，從而證明版權的持有人。本論文提出一種以基因演算法為基礎的強健式音訊浮水印，方法主要利用MPEG聽覺模型以及基因演算法，找尋最適合的水印能量大小，避免加入水印後，不會為載體訊號品質帶來太大影響，也兼顧其抗攻擊之強軔性。

The topic of intellectual property right protection has been noticed. Nowadays, due to the rapid development of computer multimedia technology, people can easily produced, process and retrieve audio media by using digital devices. Therefore, it induced two problems, the security of digital message transfer and the intellectual property right protection of the work in digital form. This essay proposed a method of embedding a watermark, hidden message about the copyright. Whenever a quarrel about the owner of copyright occurs, we can extract the watermark and prove the ownership. This method adopts acoustic models and genetic algorithms to determine the proper scaling of watermark energy such that the quality of the original signal and robustness of the watermark are still preserved.