植基於邊緣偵測法及鏡射三角定位概念之資訊隱藏技術

現行資訊隱藏方法中有一類技術是利用影像插值概念，將影像放大成為預測影像，其作法是在兩兩像素間插入一個預測像素，並且將機密訊息嵌入預測像素中。此類技術能夠巧妙的在影像放大的同時，將機密訊息嵌入，達到高藏量的目的，而且訊息取出後，再將插入的像素刪除即可還原原始影像。Jung和Yoo學者於2009年提出以鄰近像素值平均值做為預測值的方法，提出影像內插隱藏技術，其後Lee和Huang學者改良Jung和Yoo的方法，利用相鄰像素插值法提高藏入量。然而，Lee和Huang的方法會有很嚴重的鋸齒狀效應，因此本篇論文將利用可以減少邊緣效應的中間邊緣偵測法（Median Edge Detector，MED）及鏡射三角形方位法，提出一個改善Lee和Huang學者影像品質的技術。實驗結果顯示，所提方法能有效改善Lee和Huang學者的影像品質，藏入量也較其他學者高。

One of the popular strategy which used in information hiding is interpolation method. The interpolation-based hiding method expands a cover image into a larger matrix as a prediction image and hides the secret message in the predicted pixel. In 2009, Jung and Yoo proposed an interpolation based approach which used the average value of the neighboring pixel as the prediction pixel. In 2012, Lee and Huang improved Jung and Yoo's scheme with high hiding payload. In this paper, an improving method is proposed to enhance Lee and Huang's method for reducing the jagged phenomena. The proposed scheme uses median edge detector and the triangular mirroring technique to decrease the image distortion and increasing hiding capacity.