運用拉格朗日插值法之可逆式資訊隱藏法

A Reversible Data Hiding Exploiting Lagrange Interpolation

高承佑(Cheng-Yu Kao)

指導教授 ： 李金鳳

朝陽科技大學/資訊學院/資訊管理系/碩士(2014年)

摘要

隨著網際網路的進步，人們會透過網路的方式傳遞訊息，但在傳遞的過程中，會被有心人士從中進行攔截並惡意破壞或是竊取重要的訊息，因此如何提升傳遞中訊息的安全性，逐漸被網路上的使用者重視。也促使越來越多的學者致力於資訊隱藏的研究。可逆式資訊隱藏技術之研究於近年來開始蓬勃發展，其為資訊隱藏技術的一個分支，將機密訊息透過偽裝方式來傳達，來避開不當的存取或攻擊。此類技術不僅可提供機密資料安全的溝通管道，而且還能達到載體影像無失真回復的目標，這個特性使得在某些特殊專業領域的應用上相當需要與適合，例如：醫療、軍事、衛星觀測等用途，因為這些領域的影像資訊皆要求較高的精密度，不容許絲毫地誤差和失真來影響專業上的判斷。
預測誤差直方位移修正法是一種可逆式的方法，誤差愈接近其峰值點的量就會愈多就可增加訊息藏入量。因此如何產生好的預測結果一直都是研究重點。本篇論文提出運用拉格朗日插值法之四階段可逆式資訊隱藏法，以四階段模式在奇偶像素上交錯運用拉格朗日插值法來產生預測影像，運用預測直方位移修正法將機密訊息藏入預測影像與原始影像的差值中，所得到的偽裝影像不僅有很高的藏量，其平均藏量0.39bpp，且可維持一定的影像品質，其平均44dB，此外本文亦針對目前已提出有關於預測誤差直方位移修正法進行介紹與分析，從資訊的負載量、視覺影像品質與各學者的研究結果進行比較，並探討未來相關領域可以延伸的部分。
實驗結果表明本篇方法在Lena影像有43.97dB和0.38bpp的嵌入量，在多層藏入到第五層時，嵌入量為0.98 bpp且仍可以保持31.14dB的影像品質，代表本方法比Ni et al和 Tai et al的方法較佳。

關鍵字

預測直方位移修正法 ； 直方位移修正法 ； 可逆式資訊隱藏 ； 拉格朗日插值法

並列摘要

The advancement of Internet encourages network messaging, but important information could be intercepted and sabotaged or stolen in the course of delivery. Therefore, there has been growing attention on how to enhance the security of message transmission by increasing dedication to the study of information hiding. Reversible data hiding technology not only provides a safe communication channel for confidential information but also achieves lossless carrier on image recovery. This functionality makes it extremely suitable for certain professional areas like medical, military, and satellite observations, for high precision image information requirement in which areas the slightest error or distortion is not allowed.
The Histogram Shifting Modification Method for prediction error is a reversible method. The closer the prediction, the more numbers of peak points; and the higher embedding capacity it will have. Therefore, finding solutions to make better predictions has always been the focus of research. This thesis proposes the use of a four-phase reversible data hiding method to produce prediction images with Lagrange Interpolation on odd and even pixels and to hide confidential information into the difference between the predicted image and the original image by the means of the prediction histogram shifting modification method. The stego-image has not only a high payload, averagely 0.39 bpp, but also a certain image quality, 44 dB on average. In addition, the analysis of proposed method for prediction error is also carried out in the experiments of this thesis. The performance in terms of payload, visual image quality for single-layer or multiple-layer embedding are performed and compared with the research results of different scholars. We also explore the possible future extension to the related fields.
The experimental results show that this method has an embedding capacity of 0.38 bpp and 43.97 dB of image quality in Lena images. On the fifth-layer embedding, the embedding capacity is 0.98 bpp and the image quality is still 31.14 dB, indicating that this method is better than those of other proposed methods.

並列關鍵字

Histogram shifting ； Reversible data hiding ； Lagrange interpolation ； Prediction Histogram-shifting

參考文獻

[2]曾顯文、蔡逸民(2011)，"階層式預測之可逆式資訊隱藏技術" International Journal of Advanced Information Technologies(IJAIT), Vol. 5, No. 1, pp. 73-90.

[3]C. C. Chang, J. Y. Hsiao, and C. S. Chan(2003), “Finding Optimal Least Significant-bit Substitution in Image Hiding by Dynamic Programming Strategy,” Pattern Recognition, Vol. 36, No. 7, pp. 1583-1595.

[4]M. U. Celik, G. Sharma, A.M. Tekalp and E. Saber(2005), “Lossless Generalized-LSB Data Embedding,” IEEE Transactions on Image Processing, Vol.14, No. 2, pp. 253-266.

[5]C. K. Chan, L. M. Cheng(2004), “Hiding Data in Images by Simple LSB Substitution,” Pattern Recognition, Vol. 37, No. 3, pp. 469-474.

[6]W. Hong, T. S. Chen, Y. P. Chang, and C. W. Shiu(2010), “A High Capacity Reversible Data Hiding Scheme using Orthogonal Projection and Prediction Error Modification,”Signal Processing, Vol. 90, pp. 2911-2922.