標題: 使用改良的Patchwork演算法之音訊浮水印方法
Audio Watermarking Method Based on Modified Patchwork Algorithm
作者: 吳明彥
Ming-Yan Wu
楊 武
Wuu Yang
資訊科學與工程研究所
關鍵字: 音訊浮水印;Audio Watermarking;Patchwork Algorithm
公開日期: 2002
摘要: 由於電腦及網路科技的發展,許多資料都可以數位化後儲存,然後在網路上方便地傳送,其中又以文字、影像、聲音為主,本論文以聲音為研究對象,利用數位化的聲音資料的特性,設計良好的音訊浮水印的方法,以滿足我們要在聲音中隱藏訊息的各種需求。 本論文中,嵌入浮水印的演算法,改良自Patchwork演算法,我們使用機率統計的方法來分析嵌入函數,並且探討各個參數的意義,這幫助我們在偵測浮水印時能夠判斷聲音有沒有嵌入浮水印。另外,從實驗的結果,我們知道該如何控制嵌入函數中的參數,使聲音在嵌入浮水印後,有較強韌的浮水印,而且維持較高的聲音品質。實驗的結果顯示,本論文提出的浮水印方法對MP3壓縮和一般的信號處理都有很好的抵抗能力。
Since computer science and network are developed, many data can be digitized and stored in hard discs or compact discs. Those data include text, image, audio, etc. This paper focuses on audio data, and presents a good audio watermarking method based on modified patchwork algorithm. It can satisfy our requirement when we need to hide information in audio data. The kernel of proposed audio watermarking method is the modified patchwork algorithm. We used mathematical probability and statistics to analyze the proposed embedding function and found the significance of its parameters. This study assists us to detect the watermarks from audio data reasonably. From the experimental results, we know how to control the parameters inside the embedding function to gain more robust audio watermarks and keep high audio quality. Experimental results show that the proposed audio watermarks can be almost preserved after MP3 compression and common signal processing manipulations.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT910394052
http://hdl.handle.net/11536/70222
顯示於類別:畢業論文