Abstract

Today, multimedia (image, audio, 2D and 3D video, etc) penetrated fast and widespread into all areas of our lives. Frequent use of multimedia in almost everywhere brings some new issues and challenges about its veracity and reliability. For instance, digital media is widely used in mainstream media to present information about current events as they happen. Digital media is commonly used for analysis and archival purposes in military, finance, and health organizations. They are also accepted as forensic evidence in court of law. The use of digital media in court, military, and medical documents requires certain guarantees about its origin and veracity. Recent studies in multimedia forensics have begun to develop techniques to test the reliability and admissibility of multimedia.

In multimedia forensics, there are two main challenges. First one is analysis, and classification of specific multimedia characteristics (image/video statistics, noise, internal and/or external factors which affect the quality of multimedia) related with the acquisition device. Such an analysis makes possible to establish a link from multimedia output to its source device, brand, or model. This is shortly called source identification. The second challenge is investigating the veracity of multimedia using intrinsic (noise modeling) or extrinsic (watermarks) methods. This is simply called tamper or doctoring detection. Considering these two key issues, in this thesis, we will introduce several forensic techniques for source class (CG or Real/compact or DSLR) identification, source device (flatbed scanner and DSLR) identification, image tamper detection, and document copy detection. Each technique in this thesis will be presented and elaborated in separate chapters and substantiated with theoretical and/or experimental analysis.