在VQ壓縮域上資訊隱藏技術之研究

Research of Data Hiding Techniques in Vector Quantification Domain

莊梅君(Mei-Chun Chuang)

指導教授 ： 李金鳳

朝陽科技大學/資訊學院/資訊管理系/碩士(2010年)

摘要

由於網際網路的快速發展與數位時代的來臨，資料加密變得相當重要。為了提高資料傳輸時的安全性，許多學者紛紛提出資訊隱藏技術，應用在空間域、頻率域及壓縮域上，最早提出的方法有空間域上的最低位元取代法(Least Significant Bit Replacement)、頻率域上的離散餘弦轉換(Discrete Cosine Transform)與壓縮域上的VQ壓縮法及SMVQ壓縮法。為了減少網路頻寬的使用量與降低數位媒體的儲存量，本論文在壓縮技術方面提出改善的方法，應用VQ與SMVQ壓縮的技術，欲在壓縮域上進行機密資訊之藏密。
本篇論文在VQ壓縮域提出兩個資訊隱藏技術。由於EMD (Exploiting Modification Direction) 嵌入方法是針對像素值進行加減1之修改，因此運用EMD嵌入法將機密資訊藏在掩護影像內，可以有效地提昇偽裝影像的視覺品質。在壓縮域事先將編碼簿進行排序，所得到的索引表之編碼字也呈現鄰近值相似的特性。因此我們運用EMD嵌入法將機密訊息嵌入在VQ索引值中，亦能使嵌入後的索引表上之編碼字只變化1或-1。在運用EMD將嵌在偽裝索引表編碼字上的訊息有效地取回後，再運用排序過後的編碼簿進行解碼，可使重建後的影像保有良好的影像品質。
由於Wu等學者的區塊預測法的資訊藏量頗高，如果再度提升資訊藏量會使得影像品質變得更差，所以我們提出第二個方法為“植基於區塊預測之可逆式連鎖位移法”以提升機密資訊藏量，並在機密資訊取回後將影像還原到壓縮後的視覺品質。本方法設定一個變異數門檻值將區塊分成五個類型並採用SMVQ的概念在壓縮域上嵌入機密資訊，接著運用可逆式連鎖位移法於可藏區塊內嵌入更多的訊息。實驗證明本方法可以有效地提升資訊藏量，其偽裝影像亦保持良好的視覺品質。

關鍵字

直方圖 ； 邊緣吻合向量量化法 ； 資料壓縮 ； 資訊隱藏

並列摘要

By the rapid development of the Internet and the arrival of the digital era, the data encryption becomes more and more important. To enhance the security of data transfer, many scholars have proposed a lot of techniques of information hiding. The techniques which are applied in the spatial, frequency and compressed domain, example for Least Significant Bit Replacement (LSB Replacement) of the spatial domain, Discrete Cosine Transform (DCT) of the frequency domain, vector quantization (VQ) and side match vector quantization (SMVQ) of the compressed domain. To decrease the usage of Internet bandwidth and lower the size of multimedia files, we proposed two techniques related to VQ and SMVQ method to embed the secret message in the compressed domain.
In the research, we proposed two data hiding techniques based on the compressed domain. In Exploiting Modification Direction (EMD) scheme, the pixel is only modified as 1 to embed secret data in the cover image. Therefore, the stego-image quality presented high visual quality by the scheme. After sorting results of the codebook for the compressed domain, the codeword of index table resembles the character of neighbor others. Then our proposed scheme embedded the secret data to index table, the codeword of index table adds or subtracts one. After extracting the secret data and using the sorting codebook to decode the stego-image. The decoded image kept a good image quality.
Wu et al. proposed block prediction scheme that had high embedding capacity. If increasing the secret message, the stego-image quality will become worse. The second scheme proposed in our research is called “An embedding technique based upon block prediction and code cascading shift”. This method increased the embedding capacity, and recovered the image quality of the decoded image after extracting the secret data. The proposed scheme set the variable threshold to divide into five types and embedded the secret message by SMVQ in the compressed domain, then hiding more and more secret data by code cascading shift method in the embedded blocks. The experimental results show the embedding capacity of the proposed scheme is better than Wu et al.’s scheme. Furthermore, the stego-image kept a high image quality.

並列關鍵字

histogram ； side match vector quantization ； data compression ； steganography

參考文獻

[3] 紀岡良(2004)，低位元率向量量化編碼法之設計，碩士論文。

[5] 葉秀蓮(2007)，植基於預測編碼之可回復式資訊隱藏技術，碩士論文。

[8] C. C. Chang, The Duc Kieu, and Y. C. Chou, “Reversible information hiding for VQ indices based on locally adaptive coding,” Journal of Visual Communication and Image Representation, Vol. 20, No. 1, pp. 57-64, 2009.

[9] C. C. Chang, The Duc Kieu, and W. C. Wu, “A lossless data embedding technique by joint neighboring coding,” Pattern Recognition, Vol. 42, No. 7, pp. 1597-1603, 2009.

[10] C. C. Chang and C. Y. Lin, “Reversible steganography for VQ-compressed images using side matching and relocation,” IEEE Transactions on Information Forensics and Security, Vol. 1, No. 4, pp. 493-501, 2006.