Belgium declared a lockdown for the entire country as a result of the COVID-19 crisis on Wednesday (18 March), following the example of several European countries. A couple of hours later, Vasileios Avramelos from IDLab-MEDIA group was the first member of our faculty to publicly defend his PhD thesis in a 100% virtual online setting.
The thesis was supervised by Prof. dr. Peter Lambert and Prof. dr. ir. Glenn Van Wallendael (UGent), and the other members of the examination board consisted of Prof. dr. ir. Filip De Turck (UGent), Prof. dr. Nilesh Madhu (UGent), Prof. dr. ir. Mårten Sjöström (Mid Sweden University), Dr. ir. Jürgen Slowack (Barco) and Dr. ir. Maria Torres Vega (UGent).
Abstract
In the past decades, we have witnessed a technological evolution in digital video which has influenced our everyday lives in countless ways. Recently, we also witnessed a technological shift from watching video over a distance, to projecting virtual worlds only a few centimeters away from our eyes, i.e. virtual and augmented reality. Light field technology arises as a new field in digital imaging aiming at improving the way we perceive these virtual worlds. This technological shift naturally causes major changes in the ways we capture, compress, store, and transmit visual content. In this dissertation, we look into ways for compressing light field images and video using standardized technology. Novel approaches are proposed and tested for optimizing compression efficiency and ease of random access, two important features in multimedia compression. Moreover, we take a step further by investigating the possibility of parallel decoding for a novel model-based image and video representation approach.
Examination
During an internal meeting on February 5th 2020, the examination board thoroughly examined the candidate and decided to allow him to defend his work publicly. The public defense is intended to communicate the scientific principles of the candidate’s research to all audiences.
The jury unanimously announced that the thesis was defended successfully and awarded the candidate with the title of Doctor in Computer Science Engineering.
The presented work resulted the following publications in international journals and conference proceedings:
- Hard Real-Time, Pixel-Parallel Rendering of Light Field Videos using Steered Mixture-of-Experts
- Steered Mixture-of-Experts for Light Field Video Coding
- Highly Parallel Steered Mixture-of-Experts Rendering at Pixel-level for Image and Light Field Data
- Overview of MV-HEVC prediction structures for light field video
- Light Field Image Compression Using Versatile Video Coding (VVC)
- Random access prediction structures for light field video coding with MV-HEVC
The full thesis and presentation are available via the Ghent University publication repository biblio.
