prikaz prve stranice dokumenta Analiza metoda za povećanje slike
  Download
PDF 3.62 MB
master's thesis
Analiza metoda za povećanje slike
2023. urn:nbn:hr:166:472416
Jovanović, Ivan
University of Split
Faculty of Science
Department of Informatics

Cite this document

Jovanović, I. (2023). Analiza metoda za povećanje slike (Master's thesis). Split: University of Split, Faculty of Science. Retrieved from https://urn.nsk.hr/urn:nbn:hr:166:472416

Jovanović, Ivan. "Analiza metoda za povećanje slike." Master's thesis, University of Split, Faculty of Science, 2023. https://urn.nsk.hr/urn:nbn:hr:166:472416

Jovanović, Ivan. "Analiza metoda za povećanje slike." Master's thesis, University of Split, Faculty of Science, 2023. https://urn.nsk.hr/urn:nbn:hr:166:472416

Jovanović, I. (2023). 'Analiza metoda za povećanje slike', Master's thesis, University of Split, Faculty of Science, accessed 02 January 2025, https://urn.nsk.hr/urn:nbn:hr:166:472416

Jovanović I. Analiza metoda za povećanje slike [Master's thesis]. Split: University of Split, Faculty of Science; 2023 [cited 2025 January 02] Available at: https://urn.nsk.hr/urn:nbn:hr:166:472416

I. Jovanović, "Analiza metoda za povećanje slike", Master's thesis, University of Split, Faculty of Science, Split, 2023. Available at: https://urn.nsk.hr/urn:nbn:hr:166:472416

Please login to the repository to save this object to your list.
Metadata
TitleAnaliza metoda za povećanje slike
Title (english)Analysis of image upscaling methods
AuthorIvan Jovanović
MentorSaša Mladenović (mentor)
Committee memberSaša Mladenović (predsjednik povjerenstva)
Committee memberDivna Krpan (član povjerenstva)
Committee memberDino Nejašmić (član povjerenstva)
GranterUniversity of Split
Faculty of Science
(Department of Informatics)
Split
Defense date and country2023-09-22, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Abstract
U ovome radu predstavljene su metode za povećanje slike, krećući od tradicionalnih metoda i algoritama interpolacije, pa sve do najsuvremenijih metoda temeljenih na dubokome učenju, odnosno AI metoda za povećanje slike. Za referencu jedne od najsuvremenijih metoda odabran je i detaljnije prikazan Real-ESRGAN model, koji se kasnije u sklopu eksperimenta ovoga rada usporedio s nekoliko konkurenata različitih arhitektura. Eksperimentom se htjela provjeriti valjanost trenutnih metoda na tržištu, a istovremeno i ispitati dostupnost alata za primjenu tih metoda za proces povećanja slike lokalno, odnosno općenito pristupačnost krajnjem korisniku za osobnu upotrebu. Rezultati su zadovoljavajući i ukazuju na očigledan konstantni napredak, no kako je većina uobičajenih problema super-rezolucije još prisutna, izgleda da smo još daleko od prave super-rezolucije.
Abstract (english)
In this thesis methods for image upscaling were introduced, going from traditional methods and interpolation algorithms, all the way to state-of-the-art methods based on deep learning, respectively known as AI image upscaling methods. For reference as one of the state-of-theart methods Real-ESRGAN model was chosen, which was also presented in detail, and later on as a part of the experiment of this thesis this model was compared with few competitors with different architecture. Experiment served to check the validity of currently available methods on market, as well as to test the accessibility of tools for applying these methods for image upscaling locally, that is, in general the availability to end user for personal usage. Results are satisfactory and they indicate an obvious and constant progress, but due to common super-resolution issues still being present, it would seem we are still far away from real super-resolution.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:166:472416
Study programmeTitle: Computer Science and Technics; specializations in: major in teaching
Course: major in teaching
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra edukacije informatike i tehnike (magistar/magistra edukacije informatike i tehnike)
Type of resourceText
File originBorn digital
Access conditionsOpen access
Terms of useLicence Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) icon
Created on2023-09-26 11:27:59