Image Forensics in the Encrypted Domain.

Abstract

Encryption techniques used by forgers have thrown out a big possible challenge to forensics. Most traditional forensic tools will fail to detect the forged multimedia, which has been encrypted. Thus, image forensics in the encrypted domain (IFED) is significant. This paper presents the first introduction of IFED, encompassing its problem description, formal definition, and evaluation metrics. The focus then turns to the challenge of detecting copy-move alterations in the encrypted domain using the classic permutation encryption technique. To tackle this challenge, we introduce and develop a lightweight enhanced forensic network (LEFN) based on deep learning to facilitate automatic IFED. Extensive experiments and analyses were conducted to comprehensively validate the proposed scheme.