RT info:eu-repo/semantics/article
T1 Machine Learning-Based View Synthesis in Fourier Lightfield Microscopy
A1 Rostan, Julen
A1 Incardona, Nicolo
A1 Sánchez-Ortiga, Emilio
A1 Martínez-Corral, Manuel
A1 Latorre Carmona, Pedro
K1 Fourier lightfield microscopy
K1 View synthesis
K1 Neural radiance fields
K1 3D microscopy
K1 Informática
K1 Computer science
AB Current interest in Fourier lightfield microscopy is increasing, due to its ability to acquire3D images of thick dynamic samples. This technique is based on simultaneously capturing, in a singleshot, and with a monocular setup, a number of orthographic perspective views of 3D microscopicsamples. An essential feature of Fourier lightfield microscopy is that the number of acquired views islow, due to the trade-off relationship existing between the number of views and their correspondinglateral resolution. Therefore, it is important to have a tool for the generation of a high numberof synthesized view images, without compromising their lateral resolution. In this context weinvestigate here the use of a neural radiance field view synthesis method, originally developed for itsuse with macroscopic scenes acquired with a moving (or an array of static) digital camera(s), for itsapplication to the images acquired with a Fourier lightfield microscope. The results obtained andpresented in this paper are analyzed in terms of lateral resolution and of continuous and realisticparallax. We show that, in terms of these requirements, the proposed technique works efficiently inthe case of the epi-illumination microscopy mode.
PB MDPI
YR 2022
FD 2022-05
LK http://hdl.handle.net/10259/7338
UL http://hdl.handle.net/10259/7338
LA eng
NO This research was funded by Grant RTI2018-099041-B-I00, which is co-founded by the Ministerio de Ciencia, Innovacion y Universidades (Spain), by the European Regional Development Fund, and by Generalitat Valenciana (Spain) under Grant PROMETEO/2019/048.
DS Repositorio Institucional de la Universidad de Burgos
RD 05-may-2024