Video Forgery Detection Techniques

Video forgery detection encompasses a suite of computational methods aimed at verifying the authenticity and integrity of digital video content. With widespread availability of advanced editing tools, forged videos can incorporate splicing, object removal, frame insertion or deletion, double compression and transcoding to mislead viewers or evade legal scrutiny. Passive forensic techniques analyse inherent artefacts introduced during acquisition, compression and editing, seeking inconsistencies in pixel statistics, sensor‐noise patterns or coding structures. Active methods embed watermarks or digital signatures at capture time to facilitate later integrity checks. Traditional signal‐processing strategies examine Group of Pictures (GOP) structures, Prediction Unit distributions and correlation consistency of grey‐level values to reconstruct a video’s processing history. More recent approaches leverage deep learning—especially convolutional neural networks—to learn high-dimensional features that discriminate authentic frames from tampered ones, even under heavy compression or low resolution. Hybrid schemes combine sequential stochastic modelling with patch-based anomaly detection to localise regions of forgery, enabling robust detection of object deletion or insertion in surveillance footage. Collectively, these techniques are critical for upholding evidentiary trust in journalism, law enforcement, digital archiving and social media verification on a global scale.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Video Forgery Detection Techniques publication trend

The graph below shows the total number of articles in video forgery detection techniques across all publications each year (not limited to Nature Index journals).

Technical terms

Double compression: Re-encoding a video multiple times, producing detectable artefacts in quantisation or prediction statistics.

Group of Pictures (GOP): A sequence of video frames starting with an intra-coded frame followed by inter-coded frames, defining temporal prediction structure.

Prediction Unit (PU): A subdivision within a video frame used by modern codecs to perform intra- or inter-frame prediction, whose partition patterns reveal compression history.

Sensor Pattern Noise (SPN): Unique noise fingerprint introduced by an imaging sensor, used to trace the origin and detect forgery in video frames.

Convolutional Neural Network (CNN): A deep learning model that applies convolutional filters to extract hierarchical features from image or video data for classification or localisation.

References

1. An overview on video forensics. APSIPA Transactions on Signal and Information Processing (2012).
2. Sequential and Patch Analyses for Object Removal Video Forgery Detection and Localization. IEEE Transactions on Circuits and Systems for Video Technology (2020).
3. Video tampering localisation using features learned from authentic content. Neural Computing and Applications (2019).
4. Digital Video Tampering Detection and Localization: Review, Representations, Challenges and Algorithm. Mathematics (2022).
5. Video Inter-Frame Forgery Identification Based on Consistency of Correlation Coefficients of Gray Values. Journal of Computer and Communications (2014).
6. Detection of Double Compressed HEVC Videos Using GOP-Based PU Type Statistics. IEEE Access (2019).