Deep learning-based optimization of light field processing

As commonly acknowledged, light field technology offers a multi-dimensional extension for image processing by capturing both the intensity and direction of light rays, resulting in a more comprehensive representation of the real world and significantly enhancing human perception. However, the increased light rays’ capacity and density present challenges in information storage and transmission. Moreover, traditional display technology is limited in its ability to depict directional and depth information, and current light field display technology is inferior to high-definition displays. Therefore, in this the- sis, we propose a light-field-specific task-driven downsampling pre-processing algorithm and a depth-calibration deep learning framework designed for light field technology to ad- dress these issues. In experiment, we validate the effectiveness of our proposed methods from both quantitative and qualitative perspectives. These light field technology enhance- ment methods can be incorporated into the framework for light field processing.

Name thesis:degree_name

Ph.D. (Doctor of Philosophy)

Level thesis:degree_level

Doctoral

Discipline thesis:degree_discipline

Computer Networking and Communication Systems (UMKC)

Grantor

University of Missouri--Kansas City

Year dc:date.issued

2023

Author dc:creator

• Sun, Yangfan, 1989-

Advisor dc:contributor.advisor

• Li, Zhu

Handle dc:identifier.uri

https://hdl.handle.net/10355/98062

OAI identifier oai:identifier

oai:mospace.umsystem.edu:10355/98062