基於倒頻譜參數之數位音訊零浮水印

CEPSTRAL COEFFICIENTS-BASED ZERO-WATERWARK SCHEME FOR DIGITAL AUDIO

盧堉齊(Yu-chi Lu)

指導教授 ： 李清坤

大同大學/電機資訊學院/通訊工程研究所/碩士(2008年)

摘要

現今數位化的時代裡，資訊可以用數位的方式儲存與傳輸。透過網際網路，數位化資訊易於取得與散播，使得網路上充斥著未經授權而被竄改或複製的數位資訊。因此，為了遏止非法侵犯智慧財產權（Intellectual Property Rights），在數位資訊中嵌入數位浮水印（DigitalWatermark）就是其中一種重要的方法。傳統的浮水印技術不論是時間域或頻率域的做法，都會遭遇到兩個問題。其一是資訊加入浮水印後，不可避免地造成資訊品質降低，其二是浮水印的強健性和不可察覺性彼此無法兼顧。為解決上述問題， 若干研究學者提出零浮水印（Zero-watermark）機制。因為零浮水印的作法是將浮水印與音訊特徵結合在秘密金鑰中而非音訊本身，所以被保護的音訊與原始音訊完全相同。也因此在實務應用上，秘密金鑰需要送交第三公正機構做數位時戳認證，輔助解決多重浮水印的問題。 本論文將零浮水印的概念應用在音訊中，提出兩種基於倒頻譜參數（Cepstral Coefficients）的數位音訊零浮水印技術。第一種萃取特徵的方法是利用離散小波轉換（Discrete Wavelet Transformation, DWT）提供在頻率之多重解析度擷取音訊的低頻成分，結合語音辨識中用以取得特徵值的線性預估倒頻譜參數（Linear Predication Cepstral Coefficients, LPCC）萃取音訊低頻特徵。第二種方法是根據梅爾刻度頻率（Mel-frequency）具有強調低頻的特性，取得強化音訊低頻的梅爾刻度倒頻譜參數（Mel-frequency Cepstral Coefficients, MFCC）做為音訊特徵。分別用取得的音訊特徵和浮水印做互斥運算產生秘密金鑰。 實驗結果顯示測試音訊經過各種攻擊方式，如重新取樣、重新量化、低通濾波、MP3 壓縮、加入白雜訊、刪除取樣點等，我們提出的兩種方法仍然能有效還原出可辨識之浮水印圖像。實驗結果也顯示結合離散小波轉換與線性預估倒頻譜參數的方法萃取音訊低頻特徵，優於直接使用強化音訊低頻的梅爾刻度倒頻譜參數，而且以音訊低頻特徵建立零浮水印之效果較佳。實驗結果也符合數位浮水印的技術需求，如透明性、強健性、安全性、明確性和還原浮水印時不需要原始音訊輔助。

關鍵字

倒頻譜參數 ； 零浮水印 ； 音訊浮水印

並列摘要

According to the development of the Internet, unauthorized copying and distribution of digital data creates a disastrous problem in the protection
of intellectual property rights. The embedding of digital watermark into
multimedia content has been proposed to deal with this problem. In traditional watermarking algorithms, the insertion of watermark into the host signal inevitably introduces perceptible quality degradation and the inherent conflict between imperceptibility and robustness. Zero-watermarking technique was proposed to solve these problems successfully. This scheme extracts some essential characteristics from host signal and uses them for watermark detection instead of embedding watermark into original signal.
In this paper, two cepstrum based zero-watermarking schemes for digital audio are presented. In first method, discrete wavelet transform (DWT) and linear prediction cepstral coefficients (LPCCs) are combined to extract audio features which are used for watermark embedding and recovery. In second method, we use the mel-frequency cepstral coefficients (MFCCs) to extract audio features for watermark embedding and recovery. In actual applications, the secret key which was constructed after the zero-watermarking schemes should be assigned a time stamp by authority authentication department.
Simulation results show that our schemes can get the embedded watermark back after taking several attacks, and experiments show that the recovery performance of high sampling rate audio which used the first method is batter than which used the second method. The Simulation results satisfy the digital watermark technical properties of transparency, robustness, unambiguity, security, and blindness.

並列關鍵字

cepstral coefficients ； zero-watermark ； audio watermark

參考文獻

[3] C.-C. Chang, K.-F. Hwang, and M.-S. Hwang, “A block based digital watermarks for copy protection of images,” in Proceedings of the 5th Asia-Pacific Conference on Communications (APCC), 1999, vol. 2, pp. 977-980.

[5] L. Jing and F. Liu, “Double zero-watermarks scheme utilizing scale invariant feature transform and log-polar mapping,” in Proceedings of the IEEE International Conference on Multimedia and Expo, 2007, pp. 2118-2121.

[7] Y.-F. Hu, S.-A. Zhu, and D.-W. Zhang, “A novel zero-watermark algorithm in image subspace domain,” in Proceedings of IEEE International Conference on Control and Automation (ICCA), 2007, pp. 2744-2748.

[9] S.-K. Lee and Y.-S. Ho, “Digital audio watermarking in the cepstrum domain,” IEEE Transactions on Consumer Electronics, vol. 46, Issue 3, pp.744-750, 2000.

[10] I. J. Cox, J. Kilian, F. T. Leighton, and T. Shamoon, “Secure spread spectrum watermarking for multimedia,” IEEE Transactions on Image Processing, vol. 6, no. 12, pp. 1673-1687, 1997.