透過您的圖書館登入
IP:3.231.55.243

摘要


EMD(Exploiting Modification Direction)訊息嵌入演算法,該方法提供了高效率和高品質的藏密技術,在相同藏量下有著更高的嵌入效率,也使偽裝影像不易被察覺。但此方法有三個缺點:一、使用者不知如何設定參數(n, K)來滿足使用者在嵌入前對秘密訊息嵌入量或偽裝影像品質之需求;二、即便使用者設定一組(n, K)參數,使用者也無法事先預知此組(n, K)參數所能獲得的秘密訊息嵌入量或偽裝影像品質;三、欲達到最高嵌入效率,使用者需要嘗試錯誤法,反覆測試,方能求出最佳參數(n(下標 op), K(下標 op))。本文提出一個可以預測EMD嵌入結果的技術,有效快速的解決上述問題。實驗結果顯示,實際進行秘密訊息嵌入所獲得的實際嵌入結果與我們所提供的預期的嵌入結果僅有極小的誤差。就我們所知,我們所提出的預測技術是文獻上的首創;我們的方法能在使用者輸入一組參數(n, K)後,即可建議最佳參數(n(下標 op), K(下標 op)),達到最高嵌入效率,解決使用者不易設定參數與嘗試錯誤法之問題;我們的方法能先行預測秘密訊息嵌入量或偽裝影像品質,無需實際嵌入秘密訊息。總結本文,我們的技術已成功的解決EMD嵌入演算法現存的三個問題;我們的方法兼具預測準確與執行快速之優勢,能擴展藏密學之實際應用範疇。

並列摘要


Exploiting Modification Direction (EMD) is an image data hiding algorithm providing a high embedding efficiency. However, the method has three drawbacks; it is difficult to obtain the desired (n, K) parameters; it is not able to obtain any results prior to conducting the real message embedding; it is difficult to derive an optimal (n, K) to produce the results that reveal the highest efficiency. This paper proposes a prediction scheme that can solve these three problems. Experimental results show that our prediction scheme provides a significant accuracy. To the best of our knowledge, our prediction scheme is the first approach in the literature that can foresee the embedding results in advance to the real embedding. Offering the advantages of substantial estimation accuracy and execution efficiency, our proposed method solves the three drawbacks of the EMD method, leading to feasible steganographic applications.

延伸閱讀