PhD defense of Anastasia KONIK on 02/21/23
PhD defense of Anastasia KONIK from TIMC GMCAO on February, the 21th at 2pm:
« Hybrid geometric self-calibration of radiological systems. »
Place: Salle Jacques Cartier, Maison des Langues et des Cultures, 1141 Avenue Centrale, Domaine universitaire, Gières.
- Laurent DESBAT, Professeur des Universités, Université Grenoble Alpes, Supervisor
- Voichita MAXIM, Maîtresse de conférences HDR, INSA Lyon, Reporter
- Thomas RODET, Professeur des Universités, ENS Paris, Reporter
- Valérie PERRIER, Professeure des Universités, Grenoble INP, Examiner
- Charles SOUSSEN, Professeur des Universités, CENTRALE SUPELEC, Examiner
tomography, data consistency conditions, calibration, distribution
In this work, we concentrate on self-calibration of X-ray systems. By self-calibration, we consider the situation when we need to define parameters of X-ray projection models with markers in a calibration system of unknown geometry or without markers. We consider few classical X-ray models. Firstly, the 3D cone-beam model with divergent beams and with the general source trajectory is calibrated with the bundle adjustment method. It’s shown theoretically in the 3D cone-beam geometry that any system with such an integral model cannot be geometrically calibrated better then up to a similarity transformation. Secondly, for the 2D parallel geometry with parallel beams and the 2D fan-beam geometry with divergent beams and sources on a line we propose a geometric calibration based on data consistency conditions (DCC) on distributions. In this case, we extend the known DCC from functions to distributions. We model markers with Dirac distributions and construct new analytical procedures to calibrate using special calibration cages. Lastly, by the analogy with the 2D case, we construct similar calibration procedures for the cases of the cone-beam with sources on a line and the cone-beam with sources in the plane parallel to the detector plane. We present numerical simulations in each case.