Back to search

Institutional Repository of Vilnius University

Skirtingų architektūrų neuroninių tinklų, skirtų terahercinių vaizdų apdorojimui kūrimas ir optimizavimas /

Abstract

dc:description

Terahertz (THz) images are often plagued with the problem of low spatial resolution due to the limited hardware capabilities and long imaging times. In order to solve this problem, various image processing methods are used. Machine learning models have recently made various breakthroughs in solving computer vision tasks and have been also used in THz imaging. The goal of this work is to create a few neural network models, which would improve the resolution and bring back the missing details from the low resolution THz images. To achieve this task, three convolutional neural network (CNN) models – SRGAN, SRResNet and DeblurGan were trained on simulated data. Using the simulated data these models achieved 21,515 dB, 19,271 dB and 20,657 dB PSNR values, respectively. On real THz image data SRGAN and DeblurGan performed similarly and did manage to restore certain details from the test images, performing better than SRResNet model. However, these models are extremely sensitive to noise due to the strong mapping ability of the CNN: these methods tend to over-restore details and introduce severe artifacts in the final image.

Degree

thesis:*
Grantor dc:publisher
Institutional Repository of Vilnius University
Year dc:date
2023

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Šilingaitė, Ugnė,

Rights

dc:rights
Statement dc:rights
  • info:eu-repo/semantics/openAccess
Language dc:language
lit

Identifiers

dc:identifier.*

Chain of custody

source
Harvested from
Vilnius University
Base URL
epublications.vu.lt/oai
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Šilingaitė, Ugnė,. Skirtingų architektūrų neuroninių tinklų, skirtų terahercinių vaizdų apdorojimui kūrimas ir optimizavimas /. Institutional Repository of Vilnius University, 2023. https://repository.vu.lt/VU:ELABAETD192833338&prefLang=en_US