植基於雙影像之可逆式藏入法使用中間對折策略與區間量表

Lu等學者於 2015年提出使用中間對折策略之可逆式雙影像藏入法，該方法首先設定一個固定的數值 k來決定每個像素之藏入量，接著將藏入訊息切割出來，並從2進制字元轉為10進制符號來進行嵌入，透過使用中間對折策略，減少影像的失真，而後將對折後數值平均藏入兩張偽裝影像中。然而，Lu等人的方法不管影像影像特性為何，所有像素皆藏入k個位元，然而不同影像特性不同，應彈性調整像素藏入位元數。因此，本論文加入區間量表變動藏入位元數k值。所提方法先將影像切割成1×2不重疊的區塊，接著計算區塊內相鄰像素值的差異值，並帶入區間量表，以求得每個區塊可藏入之數值，為了使區間量表的設定更加簡單，在本篇方法加入預處理步驟從區間差異中找出峰值自動依照規則設定k值使區間量表能夠自動產生。實驗結果顯示本論文所提之方法能夠在一定影像品質下，同時藏入更多訊息且擁有很高的結構相似性。

關鍵字

可逆式資訊隱藏；雙影像；中間對折策略；區間量表；自適應

並列摘要

Lu et al. proposed a dual-images-based reversible data hiding method using center folding strategy in 2015. The method used a fixed value called k , to decide the hiding capacity in each pixel. The secret data was transformed from binary string to the decimal system, the secret data has applying center folding strategy to reduce image distortion. Then, the folded secret data was embedded into two stego images. The hiding capacity of Lu et al.’s scheme is limited. To make it flexible, this thesis applies interval scale strategy to dynamically change k to increase hiding payload and structural similarity. First at all, the proposed scheme divides the input image into several non-overlap blocks sized 1×2 . Then, calculates the difference between two pixels in each block. The interval scale table is used to determine the k of the block to control the image quality and to obtain higher hiding capacity. To make setting interval scale more easier, the proposed scheme add a pre-process stage to calculate all difference in each bloack, and find peak to auto create interval scale by rules. The experimental results shows that the proposed scheme can get higher capacity and still have good image quilaty and structural similarity.