資訊隱藏在影像驗證之應用

Abstract

電腦網路和資訊數位化帶來了易散佈和易修改的便利性，但數位多媒體資料在使用上的安全問題也隨之而來。例如:在一個開放式的環境傳送多媒體資料，資料在傳送過程中十分容易被擷取及竄改，若使用者得到了一張已被竄改後的影像，且使用者無原圖可供比對，則多媒體資料的可靠性十分難以判定。因此，如何確認多媒體資料的可靠性及完整性，便顯得十分重要。近來所發展之數位浮水印技術，正可有效解決多媒體資料之可靠性與完整性的驗証問題。 在本論文之第一部份中，我們提出了可復原之易碎式數位浮水印技術，若影像受到各式的竄改或破壞，或在傳送過程中遺失了部份影像資料，本技術能在不需原圖的輔助下，正確定位影像被更改的部份，並利用隱藏的浮水印將其還原。我們採取JPEG-like影像壓縮技術處理後的影像為浮水印的原型(prototype)，並使用同構轉換（Toral Automorphism）將浮水印散佈在影像中，並利用不可逆之雜湊函式及非對稱金鑰之加密函式來增加浮水印的安全性。經由實驗的驗証，本技術能在受保護之影像被竄改後，且不需原圖輔助之情況下，仍能有效的指出影像被竄改之部份，並正確的復原影像被更改的資料。 在本論文之第二部份中，我們提出了混合式的半易碎式數位浮水印技術，能同時滿足半易碎式的三個主要特性：適性的嵌入浮水印、能抵禦JPEG壓縮及能正確的偵測影像被修改的部份。本技術主要是分析JPEG對影像造成之影響，再依影像係數的特性，以不用的嵌入方式，將浮水印嵌入影像係數中，以同時達到較佳的影像品質與能忍受JPEG壓縮所造成之失真。我們同時設計了兩個校正函式，以提高驗証結果的精準度。經由實驗的驗証，本技術以混合方式嵌入浮水印後，仍能保持高度之影像品質，不但能正確測得影像中被竄改的部份，並能正確區隔JPEG壓縮與影像增強或復原技術，對隱藏之浮水印造成的影響。Computer network and digitized information, which bring the convenience of spreading and modifying, make it much easier to acquire multimedia data. Therefore, it becomes more and more important to affirm the legal use of multimedia data to protect the security and credibility of shared data. Digital Watermarking, developing recently, has been proposed as an efficient way to authenticate the integrity of protected image. In the first part of this dissertation, we presented a multipurpose fragile watermarking technique that adopts JPEG-like compressed watermark prototypes. The prototypes among blocks are spread by toral automorphism with an asymmetric key structure. Furthermore, the watermark prototypes are disordered by a one-way hashing function and encrypted by a public-key encryption function to increase the security of watermarks. Finally, the watermarks are embedded to protect the integrity of images. The experiments show that the proposed scheme can correctly localize tampered areas and remedy them to the extent of JPEG-compression quality even when 50% of the image is cropped. In the second part of this dissertation, we proposed an adaptive watermarking technique that is robust to JPEG compression and sensitive to malicious manipulations. The watermarking technique includes three major functions: generating the content-based watermarks, hybrid watermarking, and calibrating. These watermarks are embedded using various strategies according to the characteristics of the DCT coefficients. Finally, two calibrating functions are used to increase the accuracy of authentication. Experimental results from the proposed scheme demonstrate that it is able to blindly localize tampered areas, robust to JPEG compression, and sensitive to malicious manipulations, including the counterfeiting attack.