Introduction to Homomorphic Cryptosystems

2013-02-26
Held by Slobodan Petrovic

Fake resistant fingerprint biometrics / (Optical Coherence Tomography for Fingerprint Sensing)

2013-02-12
Held by Ctiraz Sousedik. Spoofing techniques, latex/silicon, software vs hardware, destroying own print, depth scan, find peak outer and inner finger print and also the sweat spirals, power minimization, curse of dimensionality, must be within 1-2 seconds.

Linear consistency test of stream ciphers

2013-02-05
Held by Slobodan Petrovic. By using a stream cipher to clock another stream cipher, edit distance can be used to reduce brute force complexity to only one stream cipher...

Ear Recognition

2013-01-29
Held by Anika Pflug. How to detect ears in a 2D image, how it is difficult because of different angles.

Sample quality in biometrics

2013-01-22
Held by Christoph Bush. How to measure the quality of a biometric sample, like fingerprint or iris. The goal was to determine at sampling time whether the sample is good enough, if it resembles the "real thing" close enough. The framework was NIST fingerprint image quality (NFIQ) and the result is supposed to be a measure integer between 0-100 where 100 is the best quality. In finger printing detecting minutiae is the goal. It is used in eID and passports (also to avoid giving away multiple passes to the same person). A question was raised whether police had access to the same registers as eID and passports. The answer was probably no.. The method explained for determining a scale from an image of a finger print was based on machine learning (neural networks) and he also explained using frequency analysis where a peak in the histogram (between a lower and upper value) would predict a good sample.

Quantum Cryptology

2013-01-15
Øystein Marøy, a PhD candidate from NTNU held a presentation on his work on quantum cryptology, distribution of keys to be precise. The main problem proposed is that quantum computers can solve factorization problems in polynomial time (versus exponential time) and thus the need for stronger cryptographic methods. Quantum bits differ from normal bits in that they, when created are not yet determined to be either 0 or 1, but are determined when observed/measured. He showed a protocol for transmitting a key so that an eavesdropper can be discovered. He talked about two different sets of bits illustrated by 90, 45, 0 and -45 degrees shifts. Had to go before the presentation was complete because of another class...