Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/19974
標題: 具變動控制性之資訊隱藏演算法
A Study of Variation Controllability Data Hiding Algorithms
作者: 侯智皓
Hou, Chih-Hao
關鍵字: 資訊隱藏;Data hiding;像素群策略;三階編碼;訊息機率;期望變動量;可預測性;視覺差異評估;偽裝學;pixel clustering;triplet coding;message probability;expected mean squared error;predictable;visual difference assessment;steganography
出版社: 資訊科學與工程學系所
引用: [Bend1996] Bender, W., Gruhl, D., Morimoto, N., and Lu, A., “Techniques for Data Hiding,” IBM Systems Journal, Vol. 35, No. 3-4, pp. 313-336 (1996). [Borg2008] Borges, P., Mayer, J., and Izquierdo, E., “Robust and Transparent Color Modulation for Text Data Hiding,” IEEE Transactions on Multimedia, Vol. 10, No. 8, pp. 1479-1489 (2008). [Chan2001] Chan, C.K. and Cheng, L.M., “Improved Hiding Data in Images by Optimal Moderately-Significant-Bit Replacement,” IET Electronics Letters, Vol. 37, No. 16, pp. 1017-1018 (2001). [Chan2004] Chan, C.K. and Cheng, L.M., “Hiding Data in Images by Simple LSB Substitution,” Pattern Recognition, Vol. 37, No. 3, pp. 469-474 (2004). [Chao2009] Chao, M.W., Lin, C.H., Yu, C.W., and Lee, T.Y., “A High Capacity 3D Steganography Algorithm,” IEEE Transactions on Visualization and Computer Graphics, Vol. 15, No. 2, pp. 274-284 (2009). [Chen2009] Cheng, Y.M. and Wang, C.M., “A Novel Approach to Steganography in High - Dynamic-Range Images,” IEEE Multimedia, Vol. 16, No. 3, pp. 70-80 (2009). [Ched2010] Cheddad, A., Condell, J., Curran, K., and McKevitt, P., “Digital Image Steganography: Survey and Analysis of Current Methods,” Signal Processing, Vol. 90, No. 3, pp. 727-752 (2010). [Cox2008] Cox, I. J., Miller, M. L., Bloom, J. A., Fridrich, J., and Kalker, T., Digital Watermarking and Steganography, 2nd Edition, Morgan Kaufmann, San Francisco (2008). [Egiaz2006] Egiazarian, K., Astola, J., Ponomarenko, N., Lukin, V., Battisti, F., and Carli, M., “Two New Full-reference Quality Metrics Based on HVS,” Proceedings of the Second International Workshop on Video Processing and Quality Metrics for Consumer Electronics (VPQM), Scottsdale, Arizona, USA (2006). [Frid2001-a] Fridrich, J., Goljan, M., and Du, R., “Detecting LSB Steganography in Color and Gray-scale Images,” IEEE Multimedia, Vol. 8, No. 4, pp. 22-28 (2001). [Frid2001-b] Fridrich, J., Goljan, M., and Du, R., “Reliable Detection of LSB Steganography in Color and Grayscale Images,” Proceedings of the ACM Workshop on Multimedia and Security (MM&Sec’01), ACM New York, USA, pp. 27-30 (2001). [Hsu2012] Hsu, F.H., Hou, C.H., and Wang, C.M., “A Data Hiding Algorithm Using the Weighted Modulus Technique,” Journal of Engineering, National Chung Hsing University, Vol. 23, No. 1, pp. 27-50 (2012). [Huan2009] Huang, N.C., Li, M.T., and Wang, C.M., “Toward Optimal embedding Capacity for Permutation Steganography,” IEEE Signal Processing Letters, Vol. 16, No. 9, pp. 802-805 (2009). [Hu2011] Hu, L.C., Lin, Y.J., Liu, Y.C., Chang, C.H., Chen, W.S., Wu, K.C., and Wang, C.M., “A Predictable Scheme for High Efficiency Data Hiding,” Journal of Engineering, National Chung Hsing University, Vol. 22, No. 1, pp. 40-54 (2011). [Jhuo2011] Jhuo, H.W., “A Predictable Data Hiding Algorithm for High Dynamic Range Images,” Master Thesis, Department of Computer science and Engineering, National Chung Hsing University, Taiwan (2011). [Kieu2011] Kieu, T.D. and Chang, C.C., “A Steganographic Scheme by Fully Exploiting Modification Directions,” Expert Systems with Applications, Vol. 38, No. 8, pp. 10648-10657 (2011). [Ko2012] Ko, Y.C. and Wang, C.M., “High Capacity and Very Low Distortion Data Embedding Algorithms for High Dynamic Range Images,” Journal of Engineering, National Chung Hsing University (Accepted 2012/11) (2012). [Liu2007] Liu, T.Y. and Tsai, W.H., “A New Steganographic Method for Data Hiding in Microsoft Word Documents by a Change Tracking Technique,” IEEE Transactions on Information Forensics and Security, Vol. 2, No. 1, pp. 24-30 (2007). [Li2008] Li, X., Zeng, T., and Yang, B., “Improvement of the Embedding Efficiency of LSB Matching by Sum and Difference Covering Set,” Proceedings of IEEE Multimedia and Expo (ICME2008), Hannover, German, pp. 209-212 (2008). [Li2009] Li, X., Yang, B., Cheng, D., and Zeng, T., “A Generalization of LSB Matching,” IEEE Signal Processing Letters, Vol. 16, No. 2, pp. 69-72 (2009). [Miel2006] Mielikainen, J., “LSB Matching Revisited,” IEEE Signal Processing Letters, Vol. 13, No. 5, pp. 285-287 (2006). [Mali2007] Malik, H.M.A., Ansari, R., and Khokhar, A.A., “Robust Data Hiding in Audio Using Allpass Filters,” IEEE Transactions on Audio, Speech, and Language Processing, Vol. 15, No. 4, pp. 1296-1304 (2007). [Peti999] Petitcolas, F.A.P., Anderson, R.J., and Kuhn, M.G., “Information Hiding – a Survey,” in Proceedings of the IEEE: Special Issue on Identification and Protection of Multimedia Content, Vol. 87, No. 7, pp. 1062-1078 (1999). [Prov2003] Provos, N. and Honeyman, P., “Hide and Seek: An Introduction to Steganography,” IEEE Security and Privacy, Vol. 1, No. 3, pp. 32-44 (2003) [Pono2007] Ponomarenko, N., Silvestri, F., Egiazarian, K., Carli, M., Astola, J., and Lukin, V., “On Between-coefficient Contrast Masking of DCT Basis Functions,” Proceedings of the Third International Workshop on Video Processing and Quality Metrics for Consumer Electronics (VPQM), Scottsdale, Arizona, USA (2007). [Stal2010] Stallings, W. Cryptography and Network Security: Principles and Practice, 5th Edition, Prentice Hall, New Jersey, USA (2010). [Wang2002] Wang, Z. and Bovik, A.C., “A Universal Image Quality Index,” IEEE Signal Processing Letters, Vol. 9, No. 3, pp. 81-84 (2002). [Wang2004] Wang, Z., Bovik, A.C., Sheikh, H.R., and Simoncelli, E.P., “Image Quality Assessment: From Error Visibility to Structural Similarity,” IEEE Transaction on Image Processing, Vol. 13, No. 4, pp. 600-612 (2004). [Wong2009] Wong, K., Tanaka, K., Takagi, K., and Nakajima, Y., “Complete Video Quality-Preserving Data Hiding,” IEEE Transactions on Circuits and Systems for Video Technology, Vol. 19, No. 10, pp. 1499-1512 (2009). [Wang2011] Wang, Z. and Li, Q., “Information Content Weighting for Perceptual Image Quality Assessment,” IEEE Transactions on Image Processing, Vol. 20, No. 5, pp. 1185-1198 (2011). [Wu2012] Wu, N.I., Wu, K.C., and Wang C.M., “Exploring Pixel-value Differencing and Base Decomposition for Low Distortion Data Embedding,” Applied Soft Computing, Vol. 12, No. 2, pp. 942-960 (2012). [Yu2011] Yu, C.M., Wu, K.C., and Wang, C.M., “A Distortion-free Data Hiding Scheme for High Dynamic Range Images,” Displays, Vol. 32, No. 5, pp. 225-236 (2011). [Zhan2006] Zhang, X. and Wang, S., “Efficient Steganographic Embedding by Exploiting Modification Direction,” IEEE Communications Letters, Vol. 10, No. 11, pp. 781-783 (2006).
摘要: 
本文提出三個以多像素群為基礎,且以影像為承載媒體的資訊隱藏演算法。我們提出的第一個演算法為「像素群策略的資訊隱藏演算法」(CPCA),具有四個優異的特性。第一,我們的演算法具備一般化特性。CPCA演算法導入藏密參數(N, K, V),其中N值規範單一像素群內含之像素個數,K值管理單一像素群嵌入秘密訊息時,像素被影響更動之個數上限,V值控制單一更動像素所允許其像素值上下變動之最大變動幅度。據此,文獻上熟知的EMD演算法即為CPCA演算法使用藏密參數(N, 1, 1)之特例。第二,CPCA方法具有預測性。我們對藏密參數(N, K, V)進行理論分析,推導出藏密參數可提供像素調整的總狀態數與期望變動量。據此理論分析,我們無需實際嵌入密訊息,即可得知掩護影像所能提供的理論訊息藏量並預測得知偽裝影像品質。第三,我們的方法具有合適性。我們採用低計算複雜度的三階編碼來充分運用總狀態數,此舉可增加額外訊息藏量,但卻不致引起更多的影像失真。第四,我們的方法具有高安全性。在實際嵌入時,演算法提供多組藏密權重及其對應之順序以供選擇,非法竊取者僅有非常小的機率能正確破解。實驗數據也證實CPCA演算法可以抵抗RS與直方圖等偽裝偵測攻擊,演算法具有高度的安全性。
我們提出的第二個演算法為「廣義型像素群策略之資訊隱藏演算法」(GPCA),來強化像素期望變動量之預測性與直方圖偽裝偵測之安全性。對於預測性,我們考量秘密訊息出現機率、嵌入演算法的特性與掩護影像具有的特徵等三種變因。演算法以矩陣的概念結合各種變因,並導出期望變動量與各變因間之函數關係。若賦予相關數值,GPCA演算法得以在尚未實際嵌入秘密訊息時,即可準確預測偽裝影像可能之失真程度與影像品質。實驗數據也證實考量這三種不同影響因素所發展出的預測機制,無需實際嵌入秘密訊息,即可精準的預測嵌入結果。我們發現若使用者給予高像素變動時,第一個CPCA演算法所產出的偽裝影像,其直方圖會顯現出異常振盪,無法抵抗直方圖偽裝偵測攻擊。對此,我們提出使用「多層像素群變更對照表」的改善策略。藉由此多層對照表,GPCA演算法使得秘密訊息與像素變動間之固定對應模式變成動態對應模式。實驗結果證實使用我們提出的多層對照表策略,確實能有效改善偽裝影像之直方圖異常振盪。量化數據顯示掩護與偽裝影像之直方圖仍維持高度相關性,大幅增強演算法抵抗直方圖偽裝偵測的能力。
最後,針對演算法的一般性與合適性,我們提出一個「通用型像素群資訊隱藏演算法」(UPCA)。在一般性方面,UPCA演算法周延考量單一像素之正向與負向變動可能不均等,故以二參數分別控制正向與負向的像素變動。據此,UPCA演算法得以包含多種相異的像素變動幅度,可以有效的因應使用者所提出的訊息藏量需求,提高演算法之一般性效用。在合適性方面,考量彩色影像亦可為承載媒體且具有不同之色彩特性,故我們加入視覺差異評估機制。對於掩護影像與偽裝影像,我們以此機制產出人眼可察覺兩者影像間具有差異之機率數值,藉此得知該彩色影像所能達到之像素變動極限,決定該張彩色影像所能提供的上限藏量。UPCA演算法不僅可以提供高嵌入藏量,且產出之偽裝影像僅有極低的機率會被人眼察覺有差異,大幅增加演算法在實質應用的合適性。
總結本文,我們提出三個資訊隱藏演算法具有四項優異特性。演算法以函數化的數學表示(N, K, Vp, Vn)並將文獻的EMD方法納為特例,故我們的演算法具備一般化的特性。我們以矩陣概念來表示影響影像品質的三種因素,藉此可在尚未實際嵌入時,即可精準預測嵌入結果,演算法具備可預測的特性。我們在訊息讀取上,採用三階編碼的方式來增加訊息藏量但卻不引起更多的影像失真。我們加入視覺差異評估機制,依據影像特性來得知每張彩色影像所能提供之藏量極限。我們的演算法具備合適性的特性。演算法除了提供多組藏密權重與其順序,更使用動態對應的多層像素變更對照表。分析顯示非法竊取者破解之機率極低,實驗數據也佐證可抵抗RS與直方圖等偽裝偵測攻擊,故演算法具有高度的安全性。綜合上述,本論文提出的演算法具備一般性、預測性、合適性與安全性,能實際擴大資訊隱藏之應用範疇。

In this study we propose three data hiding algorithms which use images as the cover media, and expand the scope of applications for data hiding.
The first algorithm is based on the concept of pixel clustering, and is referred to as a Comprehensive Pixel Clustering Algorithm for data hiding (CPCA). It contains four significant properties. First, the CPCA algorithm provides generality. We introduce three parameters (N, K, V) to cope with the different embedding strategies, where N represents the number of pixels in a cluster, K indicates the maximal number of pixels in a cluster that can be changed during the message embedding, and V controls the maximal magnitude of pixel variation for a single pixel. Thus, the EMD, a well-known data hiding scheme, becomes a special case when using our CPCA algorithm with the parameter values of (N, 1, 1). Second, the CPCA provides predictability. We analyze our algorithm and derive a mathematical function with entries of three parameters (N, K, V) in order to predict the expected mean squared error. Thus, our CPCA algorithm is able to predict the relation between the embedding capacity and the expected image quality that a cover image can offer prior to the real embedding. Third, our approach is feasible. We adopt the triplet-coding technique which not only possesses low computational complexity but also explores the full conditions of endurable pixel variations. Consequently, the CPCA algorithm is able to provide a higher embedding capacity without incurring any extra image distortion. Fourth, our scheme is secured. For message embedding, the CPCA algorithm needs to select a corresponding weight from a number of weights that are available. In addition, changing the order of any available weights does not influence the embedding results, making it difficult for an eavesdropper to determine which weight and its order are to be used for message embedding. Consequently, there is an insignificantly small probability for any eavesdropper to crack down our algorithm and unveil the secret messages. Experimental results have confirmed that the CPCA algorithm offers the capacity to defeat any RS and histogram steganalysis attacks.
We present the second algorithm, General Pixel Clustering Algorithm for data hiding (GPCA). This algorithm strengthens the predictability and security of the first algorithm. For predictability, we introduce three kinds of variables to predict the expected mean squared error which is influenced by the probability of a secret message appearance, the attributes of the embedding algorithm, and the characteristics of the cover image. With the aid of matrix multiplication the expected mean squared error becomes the function of these three variables. Consequently, once appropriate values are given, the GPCA algorithm can accurately predict the image distortion and image quality of the stego image. Experimental results confirm that the proposed prediction mechanism consideration of three variables provides prediction accuracy without actually embedding secret messages. For security purposes, we observe that employing the first algorithm to conceal a large number of secret messages leads to high pixel variation resulting in an extraordinary oscillation phenomenon which appears in the histogram of the stego image. This causes weakness of any histogram steganalysis attack. We propose a multi-level variation reference table which changes the original static mapping between the secret message and the pixel variation to dynamic mapping. Experimental results confirm the effectiveness of this approach, alleviating the oscillation appearing in the histogram of the stego image. Quantitative measuring shows that there is a high correlation between histograms of cover and stego images, significantly enhancing the capabilities of resisting histogram steganalysis attack.
Finally, we propose the third algorithm, Universal Pixel Clustering Algorithm for data hiding (UPCA), to improve the generality and feasibility of the first scheme. For generality, the UPCA algorithm takes into account asymmetrical pixel variation in the positive or negative changes using two parameters to control positive or negative pixel changes. This leads to producing a variety of patterns from pixel variations and therefore provides different embedding capacities. Consequently, the UPCA algorithm can satisfy the various demands of capacity requested by users, and greatly improve the generality of the algorithm. In terms of feasibility, considering that color images could be host images and that they may contain different characteristics, we employ a visual difference assessment mechanism. This mechanism can generate detectable probability values of visual differences between the cover and stego images. This allows us to restrict the upper boundary of the pixel variation, thus determining the maximum capacity that can be achieved in a color image. With the UPCA algorithm, we are able to produce a stego image which not only conveys high embedding capacity, but also produces low probability values of visual difference. Consequently, the visual difference assessment mechanism we introduce increases the feasibility of algorithms for real applications.
In conclusion, we propose three data hiding algorithms consisting of four valuable features. Our algorithms demonstrate generality since we launch a mathematical function expression in terms of four parameters (N, K, Vp, Vn) and include the EMD algorithm as a special case. Our scheme shows the predictability since we adopt matrix multiplication for three variables which predict stego image quality and achieve high accuracy of prediction prior to real message embedding. Our algorithms have feasibility. We adopt a triplet-coding method which increases message capacity without causing more image distortion and presents the visual difference assessment mechanism necessary to determine the maximum capacity offered by a color image. Finally, our scheme is flexible enough to include a set of embedding weights and their corresponding orders while using the multi-level variation reference table to reduce the histogram oscillation of the stego image. Analysis shows there is very little possibility that an eavesdropper can crack down our method. Experimental results demonstrate that the method is robust against RS and histogram steganalysis attacks, and accordingly, our algorithms contain high security. In summary, our algorithms demonstrate generality, predictability, feasibility and security which expand the scope of applications for data hiding.
URI: http://hdl.handle.net/11455/19974
其他識別: U0005-2401201316133900
Appears in Collections:資訊科學與工程學系所

Show full item record
 

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.