植基於向量量化壓縮之索引無損失真資訊隱藏編碼法

An Encoding Method for Index Lossless Information Hiding Based on Vector Quantization

施銘棋(Ming-Chin Shih)

指導教授 ： 李金鳳

朝陽科技大學/資訊學院/資訊管理系/碩士(2012年)

摘要

因應網路傳輸安全性的重要性，許多學者發展出將重要的機密資訊藏在媒體裡再將其傳送至接收方而不被第三方察覺。此外，資料傳輸前亦需將資料壓縮以縮短傳輸時間及減少傳輸的頻寬。常見的資料壓縮技術為向量量化壓縮法(Vector Quantization, VQ)。
許多學者運用VQ影像壓縮的過程中同時將機密資訊嵌入在索引值中，此種資訊隱藏技術稱為「壓縮藏密法」，此類方法需符合高安全隱密性、高資料壓縮率及高資訊藏量。壓縮藏密法在實際的應用上，則更需要滿足資料的可回復性。本研究提出植基於向量量化壓縮技術的無損失真資訊隱藏編碼法，以達到(1)高資料壓縮率(2)高資訊藏量及(3)可回復性之特質。本研究方法以排序後的編碼簿對影像進行壓縮。如此產生之索引表內的索引值將具有鄰近相似性，因而鄰近索引值間的差值會大量集中在零值及近於零值正負二側的範圍。針對此特點，本研究之編碼方法，當差值為零值時的使用最短的編碼，差值近於零值之編碼次之，而差值超過某個門檻值時，則以VQ方式編碼。如此可以達到高資料壓縮率的成效。本研究方法亦設計二階段的機密資訊藏入流程，在第一階段將機密資訊嵌入於編碼字串的前端後，於第二階段進一步將機密資訊嵌入於編碼的索引值中，以達到高資訊藏量的目的。此外，本研究方法的可回復性可將機密資訊取出後，索引表仍可還原至未藏密前的狀態。實驗結果證實運用本研究之方法，相較於2010年Wang等學者提出之方法，壓縮位元率平均減少了4.6%，能有效的增加資料壓縮率，並有效增加1.10%的資訊藏量。
實驗亦運用在不同大小的區塊上，在複雜影像上當索引表切割的區塊愈大時，相較於Wang等學者方法的基準點是位於區塊左上方角落，其與鄰近索引值的差距會因區塊變大而變大，本研究方法以區塊中間索引值為樞紐點做為與其周邊索引值差距之基準，故本研究方法會有較好的壓縮效果而且壓縮位元率的增長相較穩定。

關鍵字

無損失真資訊隱藏技術 ； 壓縮位元率 ； 資訊藏量 ； 影像處理 ； 向量量化壓縮法

並列摘要

Due to the importance of security of the network transmission, many scholars have developed technologies to conceal secret data in other seemingly innocent media such as images or text contents so that secret messages can be sent over a public channel to the recipient without awareness of third-party. In general, media will be compressed for shortening the transmission time and reducing the needs of transmission bandwidth.
In an attempt to increase the embedding capacity of data hiding and to reduce the transmission time when the hidden message is transmitted on the network, many scholars have proposed the data hiding schemes based on compression technique. This thesis proposes an encoding method for index lossless information hiding based on Vector Quantization to meet the requirements of (1) high image compression ratio, (2) high embedding capacity, and (3) VQ index recovery. Four embedding and encoding strategies are designed to improve upon the scheme of Wang et al. Instead of using one-bit flag, we adopt a flag of two bits to indicate which strategy is applied to hide secret data into VQ index values. We employ the index clusters with localization to enhance the similarity of adjacent indices and to improve the compressing codestream arrangement for reducing the bit rates of final codestream and increasing embedding capacity. First we rearrange the order of the codewords in a codebook by using codeword means. Sorted codebook increases the index localization within an index table. Then we partition the index table into non-overlapping index blocks and keep each index center of a block as a pivot point which is unchanged during the embedding and encoding procedure. Secret data are hidden within the differences between each VQ index values and the pivot. It brings up a considerable degree of index concentration in terms of block-wise view. Therefore, there is a large quantity of zeros or near zeros which are the difference values between the neighboring indices and their corresponding pivot point. Taking full advantages of the large quantity of zero or near zero differences, the embedding and encoding strategies can significantly reduce the length of codes; thereby decreasing the bit rates and increasing embedding capacity. This scheme also holds the index recovery. After secret data have been extracted, it can restore the index table completely. The experimental results show under the same embedding capacity at the first phase, the proposed scheme has lower bit rate compared with that of Wang et al.’s scheme and the gain for bit rate is 4.6% on average. With the secret embedding of second phase, the proposed scheme also gains in payload at 1.09% on average without increasing any compression ratio.

並列關鍵字

Bit rate ； Payload ； Lossless data hiding ； Vector Quantization ； Image process

參考文獻

[19] N. I. Wu, K. C. Fu, and C. M. Wang (2010), “A Novel Data Hiding Method for Grayscale Images Based on Pixel-value Differencing and Modulus Function,” Journal of Internet Technology, vol. 11, no.7, pp.1071-1082.

[24] W. C. Hsieh, N. I. Wu, and C. M. Wang (2010), “A Novel Data Hiding Scheme for Binary Images with Low Distortion,” Journal of Internet Technology, vol.11, no. 7, pp. 1057-1070.

[4] A. S. R. Rivest and L. Adleman (1978), “A Method for Obtaining Digital Signatures and Public-key Cryptosystems,” Commutations of the ACM, vol. 21, pp. 120-126.

[5] C. C. Chang, C. C. Lin, C. S. Tseng, and W.L. Tai (2007), “Reversible Hiding in DCT-Based Compressed Images,” Information Sciences, vol. 177, no. 13, pp. 2768-2786.

[6] C. C. Chang, W. C. Wu, and Y. C. Hu (2007), “Lossless recovery of a VQ Index Table with Embedded Secret Data,” Journal of Visual Communication and Image Representation, vol.18, no.3, pp.207-216.