Demozaik algoritam temeljen na dubokom učenju

undergraduate thesis

• Title (english): Deep Learning Based Demosaicking Algorithm
• Author: Dajana Šeketa
• Mentor: Davor Petrinović (mentor)
• Committee member: Davor Petrinović (predsjednik povjerenstva)
• Committee member: Ana Sović Kržić (član povjerenstva)
• Committee member: Tomislav Petković (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2019-07-12, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Postoji više vrsta senzora za dobivanje fotografije u digitalnoj kameri, a u ovom radu detaljnije je opisan postupak obrade „sirove“ fotografije dobivene na senzoru prekrivenom integriranim Bayerovim filtrom. Taj postupak naziva se demozaiciranjem, a njegov rezultat demozaicirana fotografija. Postoje razne implementacije demozaik algoritama, od ugrađenih MatLab funkcija do onih temeljenih na dubokom učenju. U ovom radu provedena je usporedba kvalitete četiri takva algoritma PSNR i PIQE testovima te su vizualnom usporedbom interpoliranih fotografija prikazani najčešći artefakti koji se javljaju prilikom demozaiciranja: artefakt lažne boje i artefakt resa.

Abstract (english)

There are several types of built-in color sensors in digital cameras, and in this paper, a more detailed description of the „raw“ image processing procedure obtained on the sensor covered with the integrated Bayer filter is described. This process is called demosaicking, and its result is a demosaiced photograph. There are various implementations of demosaicking algorithms, from embedded MatLab functions to those based on deep learning. In this paper four such algorithms are compared with the PSNR and PIQE tests. Also, visual comparison of interpolated photographs shows the most common artifacts that occur during demosaicking: false-color and zippering artifact.

• Keywords: algoritmi demozaiciranja; Bayerov filtar; duboko učenje; regresijska mreža
• Keywords (english): demosaicking algorithm; Bayer pattern; deep learning; regression neural network
• Language: croatian
• URN:NBN: urn:nbn:hr:168:244279
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2020-01-07 22:19:10