Super resolution methods for depth estimation in light sheet light field microscopy
In this Master's Thesis we explore enhanced depth estimation in light fields acquired with microscopes. We propose a neural network architecture for the production of novel angular views. We evaluate the performance of our method by comparing the precision of depth estimation in the HCI Light F...
- Autores:
-
Madrid Wolff, Jorge Andrés
- Tipo de recurso:
- Fecha de publicación:
- 2019
- Institución:
- Universidad de los Andes
- Repositorio:
- Séneca: repositorio Uniandes
- Idioma:
- eng
- OAI Identifier:
- oai:repositorio.uniandes.edu.co:1992/44325
- Acceso en línea:
- http://hdl.handle.net/1992/44325
- Palabra clave:
- Microscopia - Técnica - Investigaciones
Microscopia fluorescente - Investigaciones
Redes neurales (Computadores) - Aplicaciones - Investigaciones
Ingeniería
- Rights
- openAccess
- License
- http://creativecommons.org/licenses/by-nc-nd/4.0/
| Summary: | In this Master's Thesis we explore enhanced depth estimation in light fields acquired with microscopes. We propose a neural network architecture for the production of novel angular views. We evaluate the performance of our method by comparing the precision of depth estimation in the HCI Light Field Benchmark of its state of the art algorithm when receiving regular vs. upsampled light fields. We demonstrate reductions in the error of depth estimation by up to 12-35 percentage points. Complementarily, we present an approach to increase angular resolution in light field microscopy by providing optical sectioning of the sample with light sheets from a digital micromirror device. We also present a Fourier optics model of pattern projection from the DMD to the sample by a tube lens and a microscope objective. |
|---|