機 密 影 像 之 分 享 與 隱 藏

Abstract

隨著網際網路的日益盛行，有助於我們在網路上傳送資料。然而在開放的網際網路上進行傳輸（或存放）時，如何安全的傳送機密文件是目前十分重要的議題。一般而言，機密文件若集中於一個資訊(information carrier)來保管，則一旦這個資訊載體遭破壞或遺失，機密亦即損失。但在另一方面，若複製多份，則被盜取破解的機率亦會增加。為能解決這樣一個不易兩全其美的問題，機密分享(secret sharing)可能是最好的解決方式。 在本篇論文中，我們將機密分享的概念使用於數位影像上。當我們對機密影像作分享(sharing)後，我們會得到n張看似雜訊的分(shadows)。為了避免引起駭客的注意，我們加入了資料隱藏(data hiding)的方法,使得每張分存看似一般普通的影像，並且經由資料隱藏後的分存資料量仍相等於未包裝前。為了達到此目標，我們提出兩種量化(quantization)法：（1）固定區塊大小量化法及（2）變動區塊大小量化法。 由於影像分存在傳輸或儲存的過程中，可能遭遇到破壞，因此，我們提出一個快速分存檢驗法，來驗證是否所蒐集到的分存均為正確無誤的。此外，我們也提出另一個具辨識偽造者並修復分存的機制;我們可將這機制與快速分存檢驗法結合，以獲得一個立即檢測並快速修復的系統。

Due to the popularity of Internet, it is convenient to transmit and receive information. However, the data are easy to get intercepted or peeped in the public environment of computer network, hence how to transmit confidential data securely becomes critical research. Traditionally, the data are stored or transmitted by only one information carrier; and hence, the secrets will disappear when the information carrier is destroyed or lost. On the other hand, if we duplicate several copies of the secrets in the storage or transmission process, then the probability of being stolen will also increase. In order to overcome this problem, “secret sharing” might be the best solution. In the thesis, we apply the secret sharing concept to digital images. After sharing the secret image, we can obtain n shadow images (looking like random noise). In order to avoid hacker’s attention, we apply “Data Hiding Procedure” to make each shadow image look like ordinary image; and the size of the shadow image after data hiding is the same as the size before data hiding. To achieve this goal, two quantization procedures are proposed in the thesis: (1) fixed block size, and (2) variable block size. Since the shadow images might be attacked by hackers during transmission or storage, we propose a “Fast Faking Detection Procedure” to detect whether all of the received shadow images are genuine to ensure the recovery secret image is correct. Another proposed procedure is called “Faker Identification and Damaged Shadows Repairing Procedure”, we can combine it with the Fast Faking Detection Procedure to obtain a real-time alarm system (to judge whether fakers exist), followed by a fast repairing system.