A comprehensive taxonomy on multimedia video forgery detection techniques: challenges and novel trends

Area of Science:

• Digital Forensics
• Multimedia Security
• Artificial Intelligence

Background:

• The proliferation of digital videos across platforms necessitates robust methods for verifying content authenticity.
• Advanced video editing tools pose a significant threat to the integrity of visual data used in various sectors.
• Distinguishing genuine video content from fabricated material is crucial for trust and evidence validation.

Purpose of the Study:

• To provide a comprehensive overview of multimedia falsification detection systems.
• To categorize and discuss recent advancements in video forgery detection research.
• To identify challenges and future research directions in digital video forensics.

Main Methods:

• Review of active and passive video manipulation detection techniques.
• Exploration of deep learning algorithms including Recurrent Neural Networks (RNN), Deep Convolutional Neural Networks (DCNN), and Adaptive Neural Networks (ANN).
• Analysis of datasets, anti-forensics strategies, compression methods, and forensic tools relevant to video analysis.

Main Results:

• A soft taxonomy and detailed survey of current research in multimedia falsification detection.
• Discussion on deepfake detection, including relevant datasets and software tools.
• Identification of key challenges and promising research avenues in video forensics.

Conclusions:

• The study consolidates essential knowledge for understanding video forgery.
• It offers a broad investigation into extracting data and detecting fraudulent video content.
• The research highlights the growing importance of deep learning in digital forensics for video authenticity.