王識揚


基本資料


  
高雄縣

   
中興大學應用數學系
台灣科技大學資訊工程所

   
旅遊,閱讀,聽音樂



 


論文研究

l           指導教授古鴻炎老師

l           中文題目:使用實數倒頻譜係數與隨機雙向邏輯之聲訊浮水印方法

l           英文題目An Audio Watermarking Method Using Real Cepstrum Coefficients and Random Bidirectional Logic

l           中文摘要:本論文提出一個在實數倒頻譜域(cepstrum domain)上,結合正負邏輯亂數序列與雜步跳躍之聲訊浮水印的藏入方法,不僅可提高藏入的位元量,並且可杜絕他人在未經授權的情況下作浮水印的取出或竄改。此外我們以追蹤波形包絡及搭配同步位元之藏入,來偵測出浮水印的藏入時間點,因而增加了浮水印抵抗同步攻擊的能力。為了驗證所提出方法的可行性,我們實際建造了一個聲訊浮水印的藏入及解出系統,然後在進行幾種常見的攻擊之後,作浮水印的解出,攻擊的方式包括MP3壓縮、低通濾波和裁減等。實驗結果顯示,本論文方法均可在察覺出音質破壞的門檻值設定下,將浮水印正確地解出。

l           英文摘要In this thesis, we propose an audio watermarking method which uses the techniques of random bidirectional logic sequence and frog-jumping on the real cepstrum domain. This method can not only increase the amount of hidden bits but also prevent extracting or modifying the watermark in an unauthorized way from occuring. In addition, by means of tracking audio envelop curve and hiding synchronization bits, we can detect the time locations where the watermarks are embedded. Hence, the ability to resist synchronization attacks is improved. In order to verify the feasibility of the proposed method, we have actually built a system to hide and extract watermarks. Then, the system is tested in extracting watermarks after some types of attacks are acted, which include MP3 compression, lowpass filtering, and cropping. The results of the tests show that our method can always extract a correct watermark under the condition that people will not perceive sound quality change for the watermark-embedded audio file.

l           研究成果