Abstract
This Thesis aims at studying and developing efficient solutions for light field coding. In this regard, this Thesis proposes a light field coding solution based on the High Efficiency Video Coding (HEVC) standard and using a non-local spatial prediction scheme, named self- similarity compensated prediction. This solution is able to exploit the inherent correlations of this new type of content to achieve high rate-distortion performance, without requiring any explicit knowledge of the particular optical acquisition setup used when acquiring the content.In addition to this, aiming at allowing faster deployment of light field applications and services in the consumer market, a scalable light field coding solution that provides backward compatibility with legacy display devices (e.g., 2D, 3D stereo, and 3D multiview) is also proposed in this Thesis. The proposed display scalable solution makes use of an efficient inter-layer prediction scheme that when combined with the self-similarity compensated prediction is able to achieve, in most of the cases, better rate-distortion performance than the non-scalable HEVC solution.
Finally, to support the richer and flexible interaction functionalities that arise in light field imaging applications, this Thesis also proposes a novel scalability concept, named Field of View (FOV) scalability, as well as a FOV scalable coding solution. The FOV scalability supports progressively richer interaction functionalities in each higher layer by hierarchically organizing the light field angular information. Moreover, two novel inter-layer coding solutions are also proposed so as to achieve high rate-distortion performance in enhancement layer coding.
| Date of Award | 2017 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Peter Schelkens (Jury) |