基於賽局理論應用於數位影像浮水印技術的資訊系統安全設計

Abstract

隨著網際網路的發達以及資訊數位化技術的進步，數位多媒體已成為我們生活不可或缺的產 物，進而智慧財產權保護亦顯得格外重要。近年來，浮水印被廣泛的研究與應用，在數位內容的智 財權保護方法中，所有權人可以將數位內容的版權資訊以浮水印方式嵌入數位內容中，當產生版權 爭議時，即可直接透過可視浮水印識別出所有權人之標籤，或透過特定機制解讀出不可視浮水印的 存在與否，以作為智財權認定上的證明依據，使數位內容的版權受到保障。 然而，一方面由於設計適用於各種情形下的強健性浮水印演算法仍有改進的空間，另一方面則 是嵌入浮水印會依據嵌入的強度不同，對影像品質產生不同的破壞程度，因此，基於人類視覺系統 所發展出的影像品質衡量指標，就對如何計算嵌入浮水印的強度、抵抗浮水印遭受攻擊的能力與維 持影像品質上取得平衡點，變成是一個非常重要的課題，而目前尚未有統一的浮水印架構系統來做 為衡量的準則。 在此為期三年的專題研究中，計畫引進賽局理論的概念來發展出一應用於數位影像浮水印技術 的資訊系統安全設計，除了能用來評估最佳嵌入浮水印的參數，亦能提高其抵抗攻擊之能力。賽局 理論乃緣於西方世界探討兩個經濟個體之間當有利害衝突時，謀略運用之學。亦即針對一群決策者 在決策時，所面臨的問題與戰略行為，所進行的一套有系統的策略式互動行為分析方法，尋求在利 害衝突下的最適因應策略，透過策略推估，獲得本身的最大勝算或利益。因此，利用賽局理論的架 構，本專題將建立浮水印的嵌入者與攻擊者間的決策模型，透過賽局理論策略分析方法，並配合本 計畫所自行發展出的浮水印演算法與各式的攻擊測試評估方式，建構出一基於賽局理論架構的強健 嵌入浮水印系統，經過影像品質衡量指標的偵測，除了能讓數位內容保有一定水準的影像品質外， 亦能擁有抵抗攻擊能力的強健性，提供決策者建構出更具安全性與強韌性之數位影像浮水印演算 法，以達到更有效的數位智財權的保護。

Due to the speedy development of Internet and progress of digital information techniques, digital contents have become indispensable products in our daily life. Lately, digital watermarking has been extensively studied and regarded as a potentially effective means for protecting copyrights of digital media. The owner can use digital watermarking techniques to embed the copyright information into the digital content. While any dispute of copyright occurs, we can extract the watermark directly or through the specific procedure to distinguish the existence of watermark for copyright protection. However, it is still necessary to study more robust watermarking algorithms for various applications. Currently, there is no unified digital content protection architecture for watermarking framework model. There are two reasons: the first, developing a ubiquitious framework under different situation is still difficult. The second, determining the optimal tradeoff between the intensity of embedded watermark and the perceptual translucence for visible watermark is not an easy task. Therefore, the visual quality metrics based on the human visual system play a critical role to decide the energy of embedding watermark, the robustness under attacks and the best image quality after embedding. In this project which lasts three years, the research team plan to leverage the concept of games theory to develop a novel game-theoretic system security design for the digital image watermarking. It can not only evaluate the parameters for the optimal embedded energy, but also increase the capability to resist attacks. Game theory was originally developed from the western world which studies the confliction of interests between two economic bodies. In other words, it is an interactive behavior analysis method for solving problems when decision makers make the strategical moves. The goal of game theory is to let us find the best strategies and earn the maximum amout of benefits. Therefore, this project will apply the game theory to develop a model between the watermark embedding and various types of attacks. After such mechanism, we expect to have the best image quality under attacks. In summary, this research will resolve the security issue of visual communication and also provide the approach for the decision to make the best strategy for digital image copyright protection.