Multi-Level Passive Video Forgery Detection based on Temporal Information and Structural Similarity Index

Video forgery has become a growing concern in recent years due to its potential to spread disinformation, misrepresent the content, or create a false narrative. With the advancement of technology, it has become easier to create realistic-looking videos that can be used to deceive viewers. Video forgery detection is the process of identifying whether a video has been manipulated or altered in any way. There are several types of inter-frame video forgeries, such as frame deletion, frame insertion, frame duplication, or frame shuffling. Frame insertion forgery is a type of video manipulation in which new frames are added to a video sequence in order to misguide the viewers or alter the content of the video. In this paper, we propose a multi-level passive video frame insertion forgery detection method. In the first level, we determine temporal characteristics between adjacent frames using temporal perceptual information measurement or Temporal Information. The z-score is calculated to detect outliers and identify forgery locations. In the second level of authentication, a structural similarity index measurement (SSIM) is calculated for inserted frames that lie between locations detected. The SSIM of inserted frames is compared with the SSIM of original video frames to verify that the inserted frames are from another video. The method is evaluated on the TDTVD dataset. The experimental results show that the proposed algorithm performs better in detecting frame insertion forgery with an accuracy of 94.44%. The proposed method is compared with existing approaches in the field of video forgery detection.