Supplementary Material for "Polarization-Guided Multimedia Anti-Counterfeiting via Reversible Embedding and Tamper Localization"

The four uploaded videos serve as supplementary materials to the current paper. Among them:

Original Video refers to the unaltered video recorded for this study, in which the speaker's lip movements correspond to the sentence "Xidian University is my favorite school."

AI Video is a synthetically manipulated version of the original video generated via AI-based facial reenactment. In this version, the speaker’s lip movements have been altered to pronounce "Xidian has many outstanding students."

Result of the Original Video shows the detection outcome of the proposed method applied to the original video. As no temporally continuous abnormal frames are detected, the video is correctly identified as authentic.

Result of the AI Video illustrates the detection result for the manipulated video. Using the proposed polarization-encoded detection method, a sequence of abnormal frames is successfully identified and highlighted, leading to a correct classification of the video as a forgery.