運用群聚交集特性為基礎之影像驗證技術

Abstract

雲端檔案服務改變數位資料儲存的方式，使得擁有者只要將檔案資料上傳至雲端，即能同步到所有電腦，讓管理更加方便且更有效率，但其資料傳送及儲存之安全性備受關注。本篇論文針對影像資料提出一種數位影像驗證技術，主要用以偵測數位影像是否曾於傳送或儲存階段遭受惡意或不法之竄改，並進而針對這些疑似竄改區域加以復原，儘量恢復原本資料的樣貌。本篇論文將利用向量量化編碼技術來取得影像資料重要的驗證資訊，並進一步地運用區塊群聚交集特性來改善竄改誤判的情況與提升復原後的影像品質。從實驗結果顯示出本篇方法在相同竄改量之下，相較於其他先前方法有較精準的錯誤偵測能力以及較高品質的影像復原能力。Cloud service changes the way of digital data storage. As long as the data owner uploads digital file to the cloud, these data will be synchronized between all computer devices. It makes management easier and more efficient. However, data security during transmission and storage is one of the biggest concerns for the data owner. This paper proposes an image authentication scheme to detect whether an image has been illegally tampered with or not and to recover these suspected tampered regions into their original states. This paper utilizes the property of grouped block intersection to avoid possibly detected misjudgments and enhance the restored image quality. According to experimental results, the proposed scheme is indeed superior to other earlier works in terms of accurate detection and high-quality data recovery.