Sequence-Aligned and Cryptographically Chained Hybrid Digests for Compression-Aware Surveillance Video Authentication

This research presents a video-authentication method that combines visual, temporal, quality, and motion information to detect video manipulation while remaining tolerant to normal compression and format changes. It can identify frame insertion, deletion, and replacement, report the affected time interval, and mark unclear cases as uncertain instead of making an unreliable decision. ECDSA signatures and hash chaining also protect the stored reference records from unauthorized modification. Evaluated on 54 videos using six-source leave-one-source-out testing, the method achieved 83.33% accuracy, 100% precision, and 80% F1-score.