無需像素擴張的視覺機密分享機制

Abstract

視覺機密分享機制（visual secret sharing scheme，簡稱VSS）的主要概念是將機密影像（secret image）拆成數張分享影像（share images），還原機密影像時不需要複雜運算與專業密碼學知識，是利用人類的視覺感官系統解密。就安全性而言，任何人都無法僅由其中單一分享影像獲知機密影像上的資訊。 在傳統視覺密碼中，為了達到資訊隱藏的目的，機密影像中的每一個像素（pixel）都會被擴張成若干個子像素（sub-pixel），此作用稱為像素擴張（pixel-expansion），暴增的影像尺寸（size）加重了網路頻寬與儲存空間的負擔。另外，影像對比（contrast）大大降低的問題亦是現階段應用技術改善的重點。 本論文提出一種新的非擴張性視覺機密分享機制，架構在傳統視覺機密分享機制的基本概念上，在加密的過程中改以區塊性的像素取代技術來產生分享影像，避免影像過度放大。此外，改良式區塊性取代技術亦可以增加色階（level）的方式彌補影像對比降低的缺陷。未來的研究將專注於發展能夠產生有意義之分享影像的機制，以避免駭客的攻擊；此外，尋找一個新的技術以提升還原後的機密影像品質，亦是未來非常具有挑戰性的研究重點。

Visual secret sharing (VSS) scheme is a cryptography technique, which protects the secret information by dividing the image into several share images. Since VSS scheme uses only the human vision system for decryption, advanced cryptographic knowledge and complex computation are not needed. Although the secret information will never be revealed from a single share image for the traditional VSS scheme, the image size of the share is substantially expanded that each pixel is mapped into a block consisting of several sub-pixels. This scheme greatly deteriorates bandwidth limitation on the network and exhausts memory space on the storage system. In addition, the degraded image quality caused by reduced contrast is normally observed in the reconstructed secret image. In this thesis, we propose a novel VSS scheme with non-expandable image size by mapping a block in a secret image into its corresponding blocks with the same size in the share images. The result shows that the image quality has been improved by applying multi-level rather than two-level scheme for classifying the blocks based on the number of black points. Future works will be concentrated on developing a scheme for hiding shares in meaningful images rather than the images composed of random dots to prevent hackers' attention. Finding new schemes for improving the image quality for the reconstructed secret image is also a challenging topic in the near future.