Metode za povećanje rezolucije slika na temelju slika niže rezolucije

master's thesis

• Title (english): Methods for Increasing Image Resolution Based on Low-Resolution Images
• Author: Bruno Kantura
• Mentor: Sven Lončarić (mentor)
• Committee member: Sven Lončarić (predsjednik povjerenstva)
• Committee member: Marko Subašić (član povjerenstva)
• Committee member: Tomislav Petković (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2019-07-10, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Povećanje rezolucije slike na temelju slika niže rezolucije izazovan je problem čije rješavanje objedinjuju algoritmi i postupci pod imenom super rezolucija. Postoji mnogo kategorija super rezolucije od kojih u zadnje vrijeme nedvojbeno prevladavaju metode temeljene na dubokom učenju. Neke značajnije metode dubokim učenjem detaljno su opisane u radu kronološkim redoslijedom njihovog nastanka. Samim time, primjećuje se i kontinuirano poboljšanje rezultata kroz evoluciju algoritama, gdje se neke metode fokusiraju na doslovnu sličnost između originalne i generirane slike (EDSR), dok se one temeljene na generativnim suparničkim mrežama fokusiraju na perceptivnu sličnost (ESRGAN). Naposljetku rada, ostvarena je implementacija prve predložene metode temeljene na dubokom učenju – SRCNN.

Abstract (english)

Increasing image resolution based on low-resolution images is an ill-posed problem, which methods and algorithms aliased as super-resolution are trying to solve. There are different categories of super-resolution, but those based on deep learning are undoubtedly the most popular nowadays. Some of those are analized in the thesis in the chronological order of their proposal, where the continuous trend of result improvement follows, whether method focuses on structural similarity (EDSR) or perceptual quality (ESRGAN). Thesis ends with an implementation of the first deep learning method SRCNN.

• Keywords: super-rezolucija; SRCNN; VDSR; EDSR; SRGAN; ESRGAN
• Keywords (english): super-resolution; SRCNN; VDSR; EDSR; SRGAN; ESRGAN
• Language: croatian
• URN:NBN: urn:nbn:hr:168:372712
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2020-01-07 18:57:41