A Steganographic Method Using MRF-Synthesized Textures as Cover Images

以MRF生成之紋理為載體影像之藏密法

張鳳如(Feng-Ju Chang)

指導教授 ： 陳朝欽

國立清華大學/電機資訊學院/資訊系統與應用研究所/碩士(2011年)

https://doi.org/10.6843/NTHU.2011.00684

摘要

With the highly developed technology, the Internet is closely combined with human life. The issue of how to protect personal privacy even commercial confidentiality is attention-getting. Steganography embed secret messages into multimedia as images, audios, videos and etc. to avoid hackers abstract or alter the important data from one’s sending documents. We adopt a grayscale statistical texture synthesizer based on MRF to generate an image of user-requested size to fit the secret message. Each of our synthesized texture images as cover images contains four gray values: 30, 100, 170, and 240. We encrypt secret messages via exponent and modulo operations. Then we partition the encrypted bit sequence of the secret message as many of 2-bit words: 00, 01, 10 and 11, thus each 2-bit word naturally corresponds to one of the four gray values 30, 100, 170 and 240. To be more secure, we adopt a circular shift technique on (30, 100, 170, 240) such that the same 2-bit words need not be embedded into the same pixel values and the security could be further ensured. Experiments are given to demonstrate our approach.

關鍵字

藏密 ； 藏密學 ； 紋理 ； MRF

並列摘要

隨著科技之高度發展，網際網路與日常生活的連結日益密切。如何保護個人資料甚至是商業機密亦日漸成為受注目之議題。藏密學是一種將機密資訊藏入多媒體資料，如影像、音訊、視訊等資料的方法。如此一來可降低引起駭客注意的可能性，避免有心人從我們所傳輸出去的資料中，竊取甚至是竄改資料。

在此論文中，我們實作了以MRF為數學模型，可製造灰階統計型紋理的生成器，藉此產生能夠滿足藏入所有機密訊息所需之影像。我們讓此論文中產生之紋理僅包含了四種灰階值「30、100、170、240」。並將以指數模數運算加密後的二元機密訊息分割成四種2-bit word「00、01、10、11」如此一來便可以對應到紋理中所含之四種灰階值。爲了增加安全性，我們加入了旋轉的技巧，讓相同的2-bit word 不會都對應到相同的灰階值去做藏密。在此論文的最末，會列出實驗結果與數據。

參考文獻

[Bend1996] W. Bender et al., “Techniques for Data Hiding,” IBM Systems J., 35(3), 313 - 336, 1996.

[Chan2001] R. Chandramouli and N. Memon, “Analysis of LSB based image

[Chan2002] C.C. Chang and T.S. Chen, “A steganographic method based upon JPEG and quantization table modification,” Information Sciences, Vol. 141, No. 1 - 2, 123 - 138, 2002

[Chan2004] C.K. Chan and L.M. Cheng, “Hiding data in images by simple LSB substitution,” Pattern Recognition, Vol. 37, No. 3, 469 - 474, 2004.

[Cros1983] G.R. Cross and A.K. Jain, “Markov Random Field Texture Models,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 5, No. 1, 25 - 39, 1983.