Multimedia, Security and Forensics
The Multimedia, Security and Forensics (MSF) Group focuses on the interplay among multimedia, security and forensics technologies, which cover research on digital watermarking and authentication, data hiding, steganography and steganalysis, multimedia content protection, biometrics, usable security involving multimedia human-computer interface, and their applications for image, video, audio and binary content. Recent work on image forensics has also attracted much attention from this group including camera identification, anomaly detection, forgery and tamper detection. Some new research directions recently developed in the group include usable security involving multimedia human-computer interface and biometrics. Since 2011 the group has also started working on some areas in cyber security and digital forensics which do not involve multimedia data, e.g. infrastructure protection, security of e-banking systems, computer forensics and mobile device forensics. The group is also performing research on media computing, machine learning, visual quality assessment, error-control coding and reconfigurable multimedia coding, which play an important role of supporting research on multimedia security and forensics.
Potential applications for digital watermarking and forensics include the usage in the court of law for identifying the origin of unlawful multimedia content, to helping witnesses to identify where the blame lie, as well as for use in the proof of ownership and subsequently copyright infringement cases.
The MSF group has currently 17 active members consisting of 6 academics, 2 postdoctoral fellows and 9 PhD students. There are also several PhD students who are co-supervised by academics of MSF group. Some of the group’s research activities have attracted international recognition by winning the prestigious IET innovation in Engineering Award in 2006 for its work on research and commercialization of digital watermarking software. The group also hosted the 8th International Workshop on Digital Watermarking (IWDW’09) at the University of Surrey in 2009. IWDW is a premium conference attended by many of the experts and researchers in the field of multimedia security.
Research Funding and Collaborations
Research within MSF has been funded by the EPSRC, EU-FP7, The Royal Society, the UK Home Office, UK IPO (Intellectual Property Office), and by industrial collaborators including Thales UK Research and Technology, Charteris plc, and Intellas UK Ltd. Some of the projects funded include error control watermarking, digital and multimedia forensics, intellectual property protection, formal methods and analysis for digital watermarking protocols. Over last few years, strong research collaboration has been established with many universities in Europe, USA and Asia.
Research activities of the MSF group can be categorized into the following main areas:
- Digital Watermarking and Authentication: robust semi-fragile and fragile watermarking, error-control watermarking, copyright protection of multimedia content, content integrity and verification, self-recovery and restoration, perceptual quality metric and assessment, print and scan analysis, formal methods and analysis of buyer-seller watermarking security protocols.
- Steganography and Steganalysis: data hiding, spatial and transform domain embedding and detection techniques, support vector machine and classification techniques, multimedia steganography (audio, image and video), binary document and text data hiding.
- Multimedia Forensics: image and video forensics, camera identification, Benford’s Law, statistical analysis, anomaly detection, forgery and tamper detection, false detection rate, glare and HDR image identification.
- Usable Security: CAPTCHAs, graphical passwords, human authentication against observation attacks, enhanced (visual) HCI for security purpose (e.g., anti-phishing or anti-malware), human factors in security economics (e.g. e-banking systems), security visualization, applications of fractals and chaos in computer security.
- Multimedia Coding and Visual Quality Assessment: recovery of missing information in DCT-transformed images and videos, reconfigurable video coding and its applications in cryptography, objective image and video quality assessment, characterization of image capturing devices and displays, multimedia wireless sensor networks.
- Cybersecurity and Digital Forensics: risk prediction, reaction tools for critical infrastructures, real-time intrusion detection strategies, STUXNET worm, multi-source information fusion, graph-based modelling and rule based approaches, security of e-banking systems, cyber crime prevention and investigation (digital forensics), mobile security and forensics.
- Digital Media Content Analysis via Machine Learning and Pattern Recognitions: the primary aim of our research in this area is to enable computers to understand, perceive, and interpret digital media content, including images and videos, like humans. Specific approaches being investigated include pattern recognitions, domain-knowledge based and semantics-based machine learning.
- Biometrics: person identification and verification; multimodal information fusion; biometric sample quality and quality-based calibration; performance evaluation, modelling, and prediction; biometric menagerie and Doddington zoo; client-specific score normalization and fusion; biometric liveness detection and spoof attack; cohort-based score normalization and fusion; and pattern recognition.
Selected Videos from SULFA
SULFA is the Surrey University Library for Forensic Analysis, an open video database maintained by the MSF Group for video forensics research. The following are four selected videos from the library. More detail can be found at SULFA web site.
Original Video 1: Street View
Forged Video 1: Street View
Original Video 2: Sitting Man
Forged Video 2: Sitting Man
Group Head: Prof Anthony TS Ho