Cross-Spectrum Ocular Identity Recognition via Deep Learning
|Title||Cross-Spectrum Ocular Identity Recognition via Deep Learning|
|Summary||Cross-Spectrum Ocular Identity Recognition via Deep Learning|
|Keywords||Deep Learning, Biometrics, Ocular Recognition, Cross-Spectrum|
|TimeFrame||Winter 2018, Spring 2019|
|References|| R. Jillela and A. Ross, "Matching face against iris images using periocular information," 2014 IEEE International Conference on Image Processing (ICIP), Paris, 2014, pp. 4997-5001.
doi: 10.1109/ICIP.2014.7026012: https://ieeexplore.ieee.org/document/7026012
P. R. Nalla and A. Kumar, "Toward More Accurate Iris Recognition Using Cross-Spectral Matching," in IEEE Transactions on Image Processing, vol. 26, no. 1, pp. 208-221, Jan. 2017. doi: 10.1109/TIP.2016.2616281: https://ieeexplore.ieee.org/document/7587438
|Prerequisites||Good knowledge of applied mathematics, signal and image processing, and machine learning. Programming skills (preferably Matlab).|
|Supervisor||Kevin Hernandez-Diaz, Fernando Alonso-Fernandez, Josef Bigun|
A lot of real-world data is spread across multiple domains. Commercial iris sensor typically capture images in NIR (near-infrared) spectrum, while datasets accessible to law enforcement have been collected in the VIS (visible light) domain. Thus, there exists a need to match NIR to VIS face images.
The goal is to develop algorithms that can compare ocular images captured in different spectrum (visible, infrared, thermal) for identity verification purposes. We propose to explore recent developments in deep learning for this purpose, such as Generative Adversarial Networks (GANs) or Autoencoders.
As this is a highly active research area right now, there is a high probability of this project leading to a research publication in a reputed conference or journal.