像素值排序藏入技術使用多預測器策略

Pixel Value Ordering Hiding Method Based on Multiple Predictors

可逆式資訊隱藏技術中像素值排序藏入方法（Pixel Value Ordering, PVO）近期很常被使用，PVO 將影像切割成數個相同大小的區塊，以區塊內的像素值進行排序，機密訊息則嵌入至像素最大值與最小值當中。PVO 的影像品質與藏入能力會因區塊的大小而不同，如果區塊太大，藏入量可能受到限制。因此，本研究以多種預測器進行改良，以多種不同預測方法提升影像品質，並且結合 Ou 和 Kim 學者提出的PVO 方法，將機密訊息藏在最大預測值或最小預測值中，達到高藏入量。從實驗結果可知所提方法利用中間邊緣偵測法（Median Edge Detect, MED）進行預測，有效減少影像失真，達到高影像品質，而使用高藏量可逆式隱寫器（High Capacity Reversible Steganography, CRS）進行預測，可提升整體藏入量。

Pixel Value Ordering (PVO) hiding scheme has been used in reversible data hiding technique more often. This kind of scheme divides an image into several blocks and orders the pixels in the same block. The secret message are concealed into the maximum and minimum values. The quality of the stego image and hiding capacity by using PVO are dependent on the size of the block. Large block will cause low hiding rate. This paper combines multiple predictors to improve the hiding capacity of PVO. The experimental results show that the proposed scheme used MED can get higher image quality. The scheme applies CRS can get higher hiding capacity.