Corneal Topography: An Emerging Biometric System for Person Authenticatio

Ok...all very quite on the threads and no doubt all of us very busy at this time of the year...I know I have been but still here. Not much to add really recently as I think most of us are just disappointed and disillusioned as to how this has (been) played out for the time being

Anyway, trust all are well & decided to start a new thread.

Have still been researching, digging and surfing and found this recently but been waiting to try find a full copy - not released yet so the best I can get is the abstract as below if not seen already. Is from the 2015 BIOSIG (Biometrics Special Interest Group) conference held in Sept this year.

http://fg-biosig.gi.de/biosig2015

For me, I have bold 4 parts just from the abstract that highlights to me how others, much better credentialed and more intelligent than me, adds weight to the viability and credibility (IMO) of this tech but also potentially raises some other questions around certain other info given or lack thereof IMO. Nice EER btw.

I ask myself though, how are these scholars, scientists etc able to produce results from the concept, obviously from their own devices, maybe using Matlab & Casia databases as per links bottom of post. That, I can't answer as yet sadly without seeing the full publication and workings.

http://ieeexplore.ieee.org/xpl/logi...ieee.org/stamp/stamp.jsp?tp=&arnumber=7314597

Corneal Topography: An Emerging Biometric System for Person Authentication

Corneal topography is a non-invasive medical imaging technique to assess the shape of the cornea in ophthalmology. In this paper we demonstrate that in addition to its health care use, corneal topography could provide valuable biometric measurements for person authentication. To extract a feature vector from these images (topographies), we propose to fit the geometry of the corneal surface with Zernike polynomials, followed by a linear discriminant analysis (LDA) of the Zernike coefficients to select the most discriminating features. The results show that the proposed method reduced the typical d-dimensional Zernike feature vector (d=36) into a much lower r-dimensional feature vector (r=3), and improved the Equal Error Rate from 2.88% to 0.96%, with the added benefit of faster computation time.

Matlab - algo type database from what I can see:

https://github.com/Matlab-Biometric-recognition/Iris-Biometric-Recognition-With-Genetic-Algorithms

Casia - Iris database:

http://biometrics.idealtest.org/dbDetailForUser.do?id=4

So...AGM upon us. I would def like to go but can't get time off at the mo...still trying to get round that though and will update if situation changes for the better

Suppose though, being a realist, don't see much changing at the AGM. Whilst be enlightening in some ways to put certain individuals under the spotlight with some straight up tough questions, I honestly believe it is pretty much a fait accompli for what they want to achieve....which btw, I still believe it will be a positive result in time after relist - IMO anyway.

Have a good one all and a safe and happy Christmas if not post b4 then.