GKAMAS Theodosios
Jump to navigation
Jump to search
Ph.D. Candidate
Url : http://www.linkedin.com/pub/theodosios-gkamas/42/7a/435 |
 |
brief CV
- Theodosios N. GKAMAS was born in Ioannina, Greece, in 1985. He was admitted at the Computer Science Department of the University of Ioannina in 2003. He received his B.Sc. diploma in Computer Science in 2008 with degree 7.32/10 “Very Good”. That year he became a postgraduate student at the same institution from where he graduated in October 2010, with degree 9.90/10 "Excellent". At this moment, he is a Ph.D. candidate and a researcher at the University of Strasbourg in France. His main research interests lie in the field of Computer Vision, Signal & Image Processing, Medical Imaging, Bio-informatics and Machine Learning with specific interests in Optical Flow and Image Registration.
Specialties
- Signal & Image Processing, Image Registration, Optical Flow, Object Tracking and Development of iPhone/iPod touch/iPad Applications.
Education
- École Doctorale MSII (ED n°269) « Mathematics, Information and Engineering Sciences »
- 2011 – 2014 (expected)
- Theme: "Image processing of diffusion magnetic resonance (diffusion MRI) : construction of statistical models to determine prognostic factors of awakening in the case of coma."
- M.Sc., Computer Science, with specialization in Technologies & Applications
- 2008 – 2010
- Degree: 9.90/10, “EXCELLENT”
- Thesis: “Optical flow estimation using spatially varying smoothing”. [pdf]
- Supervisor: Assistant Professor C. Nikou.
- B.Sc., Computer Science
- 2003 – 2008
- Degree: 7.32/10, “VERY GOOD”
- Diploma Thesis: “Signal and image registration by using generalized elastic nets”.
- Supervisor: Assistant Professor C. Nikou.
Publications
Journals
1. Variational-Bayes Optical Flow
- Journal of Mathematical Imaging and Vision
- January 2014.
- Journal of Mathematical Imaging and Vision
- Authors: Giannis Chantas, Theodosios Gkamas, Christophoros Nikou
Conferences
1. Guiding optical flow estimation using superpixels
- IEEE 17th International Conference on Digital Signal Processing (DSP '11)
- July 6, 2011, Corfu, Greece.
- IEEE 17th International Conference on Digital Signal Processing (DSP '11)
- Authors: Theodosios Gkamas, Christophoros Nikou
- Abstract: In this paper, we show how the segmentation of an image into superpixels may be used as preprocessing paradigm to improve the accuracy of the optical flow estimation in an image sequence. Superpixels play the role of accurate support masks for the integration of the optical flow equation. We employ a variation of a recently proposed optical flow algorithm relying on local image properties that are taken into account only if the involved pixels belong to the same image segment. Experimental results show that the proposed optical flow estimation scheme significantly improves the accuracy of the estimated motion field with respect to other standard methods.
2. A probabilistic formulation of the optical flow problem.
- 21st International Conference on Pattern Recognition (ICPR '12)
- November 11-15, 2012, Tsukuba, Japan.
- Authors: Theodosios Gkamas, Giannis Chantas, Christophoros Nikou
- Abstract: The Horn-Schunck (HS) optical flow method is widely employed to initialize many motion estimation algorithms. In this work, a variational Bayesian approach of the HS method is presented where the motion vectors are considered to be spatially varying Student's t-distributed unobserved random variables and the only observation available is the temporal image difference. The proposed model takes into account the residual resulting from the linearization of the brightness constancy constraint by Taylor series approximation, which is also assumed to be a spatially varying Student's t-distributed observation noise. To infer the model variables and parameters we recur to variational inference methodology leading to an expectation-maximization (EM) framework in a principled probabilistic framework where all of the model parameters are estimated automatically from the data.