差值擴張技術與支援向量迴歸於可逆式資料庫浮水印技術之應用

A Study of Reversible Database Watermarking Using Difference Expansion and Support Vector Regression

張詠南(Jung-Nan Chang)

指導教授 ： 吳憲珠

國立臺中科技大學/資訊與流通學院/資訊工程系碩士班/碩士(2012年)

https://doi.org/10.6826/NUTC.2012.00118

摘要

近年來人們常將重要的資料庫放在伺服器上，提供多人使用，使得資料庫容易被非法地竄改，因此驗證資料庫內容的正確性已成為重要的研究主題。在本論文中，主要探討應用可逆式浮水印技術於關聯式資料庫保護之研究，核心技術乃使用預測誤差的方法，將數位浮水印嵌入於保護欄位值中。本論文提出的技術利用支援向量迴歸建立資料庫中欲保護欄位的預測模型，並透過支援向量迴歸預測的數值與保護數值之間的差異嵌入碎型浮水印，達到保護關聯資料庫內容之安全性及驗證。
本論文第一個提出基於支援向量迴歸預測與差值擴張方法的資料庫浮水印技術，其中支援向量迴歸預測將資料庫的欄位投影到多維空間找到迴歸的平面，藉此預測保護欄位數值。由於資料庫中許多欄位與保護欄位間並無關連性，將會造成預測數值不準確，因此本論文的方法使用FP-tree資料探勘方式找出欲保護欄位之有關聯性欄位，藉此提高支援向量迴歸預測的準確性，並且計算預測數值與保護欄位之間的差異，經由差異值擴張法將浮水印嵌入欲保護欄位中，進而達到保護資料欄位的正確性。
在第二個方法中，改良本論文第一個提出之方法，由於差異值擴張法須透過於支援向量迴歸預測數值與保護欄位的差異值進行運算，使得保護欄位會因為差異擴張的差異值大小而影響嵌入浮水印後的數值，造成保護欄位與原始欄位數值差異過大的情形，為了改良此問題，本研究進一步提出連續區域重疊差值擴張法之嵌入浮水印技術，其中本文方法分成二個步驟，首先第一個步驟為透過保護欄位區域與預測欄位區域內數值比較大小後，產生一組具有關聯性的浮水印。第二步驟將保護欄位數值轉換成二進制，由最高有效位元(MSB)區域開始嵌入浮水印，因為保護欄位與預測欄位具有MSB之區域相近值的特性，因此嵌入浮水印後原本保護欄位區域並不會變動。本文方法之連續區域重疊差值擴張法將浮水印嵌入欲保護欄位中可改善數值修改過大的情況，進而提升保護欄位之正確性。

關鍵字

碎型浮水印技術 ； 關聯式資料庫 ； 支援向量迴歸 ； 資料探勘

並列摘要

In recent years, people usually store the important databases on the server to share the significant information among users. The sharing environment makes the database easy to be modified without authentication. Therefore, applying digital watermarking to verify the correctness of the contents in the database is necessary.
In this thesis, discussion the application of the reversible database watermarking technique and protection for the relational database. The kernel technique is the error prediction method. The protected field is embedded by the watermark information. A predictive model--- support vector regression is used to predict the protected field, and the protection of the contents of the relational database is achieved by fragile watermarking with the using of the difference between predicted value and the protected value.
The first method, a reversible fragile database watermarking technology is proposed by applying Support Vector Regression (SVR) to the relational database. SVR predicts the protected field by projecting the data fields onto the Multi-dimensional space to find regression plane. Moreover, FP-tree data mining is utilized to improve the accuracy of the prediction of support vector regression. Therefore, FP-tree data mining finds the associative fields in the database for the protected field. Watermark embedding is performed by difference expansion.
The second proposed method to improve method of first. Due to the DE technology of first method needs compute with the difference value between prediction value of SVR and original protected field. However, the embedding watermark value can be affected the DE of difference value in original protected field. Afterward, the predicted filed and original protected field has larger variation.
This thesis proposed the Continuous Region Overlap Difference Expansion (CRODE) technology to solve this problem, and the proposed method divided two steps. Firstly, create the relational watermark by comparing the protected value and the predicted value, secondly transform the protected value converted into binary, and the watermarking would be embedded orderly first from MSB region. The protected field and predicted value has the similarly value with the MSB region. Therefore, the original protected field region is not change after embedding watermarked. The proposed method of Continuous Region Overlap Difference Expansion to embed the watermarked that protected field can solve the value variation and increase the accuracy of the protected field.

並列關鍵字

Fragile watermark ； Relational database ； Support vector regression ； Data mining

參考文獻

[1] M. Alattar, “Reversible Watermark Using the Difference Expansion of a Generalized Integer Transform,” IEEE Transactions on Image Process, Vol. 13, No. 8, 2004, pp. 1147-1156.

[3] R. Agrawal, P. J. Haas, and J. Kiernan, “Watermarking Relational Data: Framework, Algorithms and Analysis,” The VLDB Journal, Vol. 12, No. 2, 2003, pp. 157-169.

[6] E. Bertino, S. Jajodia, and P. Samarati, “Database Security: Research and Practice,” Information Systems, Vol. 20, No. 7, 1995, pp. 537-556.

[7] F. Bao, R. H. Deng, B. C. Ooi, and Y. Yang, “Tailored Reversible Watermarking Schemes for Authentication of Electronic Clinical Atlas,” IEEE Transactions on Information Technology in Biomedicine, Vol. 9, No. 4, 2005, pp. 554-564.

[8] C. C. Chang and C. J. Lin, LIBSVM: a library for support vector machines, 2001, software available at http://www.csie.ntu.edu.tw/~cjlin/libsvm.