Multimedia Forensics for Detecting Forgeries

Abstract

The first part of the chapter describes some examples of multimedia forgery. Here, multimedia data, including images, audio recordings or videos, etc., are forged by any of the following operations: data removal, replacement, replication, photomontage, or computer-aided media generation. The second part presents the concept of multimedia forensics and its corresponding functions. Multimedia forensics is carried out by extracting valuable information from multimedia content and using it to identify or authenticate the origin or source of multimedia and, in the process, to detect forgeries. The third part reviews general forgery detection techniques and compares their performance. Here, existing forgery detection methods are classified into 3 groups: watermarking-based scheme, perceptual hash-based scheme, and multimedia forensic-based scheme. Each of these performs at different levels of efficiency and accuracy. The fourth part investigates multimedia forensic-based forgery detection schemes. These forensic methods are composed of special features (correlation, double compression, light, and media statistical); each performs unique functions such as duplication detection, photomontage detection and synthetic image detection. The fifth part addresses some topical and timely issues, focusing on detection accuracy, counter attacks, test bed, and video forgery, etc. The last section discusses future prospects and makes some conclusions.