建構於空間-頻率域多重浮水印技術之數位影像著作權保護與認證

Digital Image Copyright Protection and Authentication Based on Spatial - Frequency Domain with Multi -Watermarking Technique

蔡曜州(Yao-Chou Tasi)

指導教授 ： 涂世雄

中原大學/工學院/電機工程研究所/碩士(2004年)

https://doi.org/10.6840/cycu200400110

摘要

摘 要
在本論文中，我們提出一個新的數位影像著作權保護與認證系統建構於空間-頻率域之多重浮水印技術。我們提出的架構結合了數位影像處理、數位浮水印、密碼學等技術。除此之外我們更提出幾個有效的方法，進而滿足數位浮水印之需求。
數位影像處理方面，我們提出一個新的空間-頻率域(或脆弱-強健)浮水印技術。主要概念是利用位元平面分割技術，將原始影像分解成頻率域影像與空間域影像。然後分別將浮水印嵌入空間域影像及頻率域影像之中，因此便可以得到二張浮水印影像。最後再將這二張浮水印影像結合成一張浮水印影像。由上述可知，我們所提出的方法可以整合空間-頻率域(或脆弱-強健)浮水印技術的特性，達到數位影像的著作權保護與認證之效能。
數位浮水印方面，我們提出二個新的浮水印技術，分別是多重浮水印技術與修改中頻係數技術。多重浮水印技術的主要概念是對原始影像嵌入多個浮水印，以達分散攻擊之效能。然而修改中頻係數技術的主要概念是藉由係數修改，然後將浮水印嵌入在被修改的係數上
，以達萃取浮水印不需原始影像之效能。由上述可知，我們所提出方法可以增強浮水印的不可視性、強健性及萃取浮水印不需原始影像。
另一方面，密碼學的主要技術也被應用在我們的架構之中，例如秘密金匙加密系統、公開金匙加密系統、單向雜湊函數與數位簽章。利用上述之密碼學技術，讓我們的架構能更滿足在網際網路使用之需求。進一步在數位簽章方面，我們更提出利用Exclusive-OR 運算技術，使摘要的形成能夠加快。
最後我們使用一些數位影像處理技術例如JPEG壓縮、塗鴉、銳化、雜訊及裁切等攻擊，對浮水印影像做一系列攻擊測試。由實驗結果可知，我們所提出的方法對於上述之攻擊皆具有良好的抵禦能力。
在本論文中，我們主要的貢獻如下:
(1) 我們整合空間-頻率域(或脆弱-強健)浮水印技術的特性，達到數位影像的著作權保護與認證。
(2) 我們更提出多重浮水印與修改中頻係數的創新技術，以增強浮水印的不可視性、強健性及萃取浮水印不需原始影像。
(3) 我們使用秘密金匙加密系統、公開金匙加密系統、單向雜湊函數與數位簽章，進而增強浮水印的安全性、驗證浮水印影像的完整性及實現不可否認性。
(4) 我們更利用Exclusive-OR 運算技術，使摘要的形成能夠加快。

我們相信本論文的研究成果，對未來數位浮水印的研究領域上有相當的助益。

關鍵字

密碼學 ； 數位影像處理 ； 著作權保護與認證 ； 數位浮水印

並列摘要

Abstract

In this thesis, we propose a new digital image copyright protection and authentication based on spatial-frequency domain with multi- watermarking technique. Our framework combines digital image processing, digital watermarking and cryptography. Moreover, we propose several valid techniques for satisfying the requirements of digital watermarking.
In digital image processing technique, we propose a new spatial- frequency domain watermarking technique. The main concept is that the original image is decomposed into spatial and frequency images, at first. And then the spatial and frequency watermarks have embedded into spatial and frequency images, respectively. Therefore, we can get two watermarked images in spatial and frequency domains. Finally, the two watermarked images were composed into a watermarked image. Consequently, this method can integrate spatial-frequency (or called fragile-robust) watermarking properties for digital image copyright protection and authentication.
In digital watermarking technique, we propose two new techniques. For the first one, the multi-watermarking technique was used in our framework. So, a lot of watermarks have embedded in digital image that this method can disperse risks for malice attack. On contrast, in the other one, the middle frequency DCT coefficients were modified in frequency domain. The frequency watermark has embedded in modifying middle frequency DCT coefficients. Consequently, these methods can increase watermark’s imperceptibility and robustness and achieve to extract the watermark without original image.
In addition, the technique of cryptography such as secret-key cryptosystem, public-key cryptosystem, one-way hash function and digital signature are utilized in our framework. The secret-key cryptosystem was used to increase watermark security that except by the copyright owner, other people must be no way to detect or remove the watermark from the watermarked image. And then the public-key cryptosystem uses a pair of mathematically related keys to execute encryption and decryption that it can increase data security and achieve non-repudiation. Besides, the one-way hash function and digital signature were used that the copyright owner can be able to verify the watermarked image has not altered while in transmission and achieve non-repudiation. Furthermore, we use exclusive-OR operation technique that it can achieve to form a digest quickly.
Finally, we use some digital image processing techniques such as sharpening, daubing, cropping, adding noise and JPEG compression to attack watermarked image. The experimental results will be given to demonstrate that our method is effective.
The contributions of our research are as follows:
(1)We propose a spatial-frequency domain watermarking technique to form a watermarked image that it can integrate both the spatial-frequency (or called fragile-robust) watermarking properties for digital image copyright protection and authentication.
(2)We propose the multi-watermarking and modifying middle frequency DCT coefficients techniques that they can disperse attack risks, increase the imperceptibility and robustness and extract watermark without original image.
(3)We use secret-key, public-key, hash function and digital signature cryptosystems to increase watermark security, verify watermarked image integrity and achieve non-repudiation.
(4)We use exclusive-OR operation technique that it can achieve to form a digest quickly.
The results of our study in this thesis will be probably practical and efficient for digital image copyright protection and authentication in internet environment.

並列關鍵字

cryptography ； digital watermarking ； digital image processing ； copyright protection and authentication

參考文獻

[33] 賴昱宏，以多重Kerberos建構無線區域網路之全域認證，中原大學電機工程學系碩士論文，2003。

[35] 董乃仁，改良UMTS認證機制之研究，中原大學電機工程學系碩 士論文，2003。

[1] A. Kejariwal, “Watermarking,” Potentials, IEEE, Volume: 22, Issue: 4, Oct. /Nov. 2003 Pages: 37 – 40.

[3] Y.C. Fan, H.W. Tsao, “Watermarking for Intellectual Property Protection,” Electronics Letters, Volume: 39, Issue: 18, 4 Sept. 2003

Pages: 1316 – 1318.