ESR3 – Evdokia Saiti

Evdokia Saiti

Country of Origin: Greece
Host Institution: NTNU


Evdokia completed her bachelor's and master's from National & Kapodistrian University of Athens (NKUA) from 2003 until 2010. Her master thesis topic is titled "Ray Casting Visualization System of Tetrahedral Meshes implemented on hardware" (Implementation using C++, OpenGL and GLSL). A visualization system of vector fields that are defined on tetrahedral meshes was implemented, using multiple levels of detail and it was improved in order to take advantage of the performance and space savings of a GPU, thus Ray Casting was exclusively implemented on the GPU. An essential element of the implementation was the exploitation of a method achieving the intersection of a ray and a tetrahedron, and therefore a tetrahedral mesh, using Plücker coordinates.

Aspirations within projects:

Evdokia is looking forward to be part of the CHANGE project and improve her educational and research skills within its framework. She am fascinated by the trends in Computer Vision and more specific in deep-learning oriented systems. In the framework of the PhD position she is keen to examine how the so-called geometric deep learning, which is the very new trend in the area of Deep Learning, can be exploited in the case of 3D mesh registration across time and modalities. Finally, as she comes from a country with rich cultural heritage, it will be an extra honor and challenge to be part of the process of developing and designing methods especially for cultural heritage assets.

ESR3: Registration techniques for differential and multimodal data


  • Developments of registration algorithms suitable for successive surface scans of 3D objects across time without external reference points
  • Developments of  multi modal registration algorithms that can automate the tedious process of registering multi modal data (e.g., surface and mineral scans) without external reference point.

Main Supervisor:


Saiti, E., T. Theoharis (2020): An application independent review of multimodal 3D registration methods. Computers & Graphics, Vol. 91, 2020, p. 153-178, ISSN 0097-8493.