Presenter

Miss Athanasia Symeonidou - ETRO, VUB [Email]

Abstract

In recent years, three-dimensional display technology has seen huge advancements, targeting a more immersive 3D experience for a plethora of applications such as medical imaging, education, CAD, entertainment and more. However, the holy grail of three-dimensional imaging remains holography, since it is the most complete representation of a 3D scene. This is due to the fact that a hologram is a recording of both the intensity and the phase of a light wavefront representing the scene, providing us with all the necessary depth cues for full parallax three-dimensional visualization. Several disadvantages of digital holography, such as the need for optical recording and reconstruction setups and the optical aberrations can be overcome by computer-generated holography, which enables the calculation of the holographic pattern of synthetic scenes as well.

However, high-quality digital holograms raise limitations and challenges, regarding the signal processing, storage and display of holographic data. In particular, the computational complexity to calculate high resolution holograms cannot be handled by current hardware, while the contemporary holographic displays are limited, not allowing for large scale digital holograms to be displayed at a large resolution.

This thesis consists of several solutions to alleviate the aforementioned challenges. First, we present a novel algorithm for efficient generation of high-quality holograms of 3D scenes based on point cloud data. This computer-generated holography algorithm not only enables the generation of high resolution colour holograms in a fast manner, but also integrates occlusion handling and shading techniques for realistic 3D visualizations. Additionally, enabled by the above algorithm, several holographic datasets have been made publicly available, consisting of holographic data of various scenes, types and physical properties. Moreover, employing light field displays to overcome the need of an optical lab or a holographic display is proposed, by introducing a framework for rendering computer-generated and analog holograms on such displays. This framework can be accompanied by the speckle noise reduction techniques proposed, to tackle the speckle noise, which is an inherent property of holography. Last but not least, a method for the efficient generation of holographic videos is proposed, which is based on a motion compensation scheme. In this context, rigid motions of an object and their effect on the wavefield of the hologram are investigated to understand the limitations set by the holographic setup and the motion compensation algorithm integrated.

Short CV

M. Sc. in Electronic Physics, Aristotle University of Thessaloniki, Greece, 2012