Electronic Imaging, Media Watermarking, Security, and Forensics 2020

Mikael Lindstrand, gonioLabs, will give a presentation "High-entropy optically variable device characterization – Facilitating multimodal authentication and capture of deep learning data", 29 January at 11:45 am.

IS&T International Symposium on Electronic Imaging, Science and Technology, Media Watermarking, Security, and Forensics (MWSF 2020), Burlingame, CA, USA, 26-30 January, 2020.

A previously presented multidimensional high-fidelity optical instrument is able to characterize the perceptually significant features of Optically Variable Devices (OVDs). This high-entropy digital information source facilitates the adoption of algorithmic-based communication protocol, and principally new services of which two will be presented.
This detailed characterization is significant for a forensic or reference tool but overly redundant for most authentication applications. Thereby, the high-entropy full characterization ability may reside at the trusted forensic authority. Distributed in a potential hostile environment are authentication devices of desirably lower characterization capabilities optimized for authentication and operational capabilities, and low cost. This protocol resembles well with the information security principle admitting information access on a need-to-know basis. Such devices may be vital improving the OVD ratio of inspection.
In a high-security application, the redundant OVD characterization facilitates challenge-response authentication, including single-use codes, prohibiting eavesdropping reply-attacks. More significantly, an intertwined multimodal (electro-optical) communication protocol is described, in which OVD-assisted authentication cooperates with cryptographic algorithms improving e.g. the sensitive access control of electronic machine-readable travel documents, eMRTD.
The characterization method, focusing only on overt (first-line inspection accessible) OVD features, facilitates inspection method monitoring, promoting this other example of multimodal (human-instrument) services, e.g. indicating possible first-line or instrument inspection shortcomings.
A combination of high-entropy and low cost instruments may be essential capturing necessary high quality, representative and large volume data for robust response deep learning algorithms, a challenge in part due to the variation of circulation caused degradation of OVD optical performance.
These services illustrate how OVD inherent capabilities may be better captured and exploited, facilitated by the described high-entropy characterization tool.


Posted  2020-01-09  by  Mikael Lindstrand