Objektivne mjere za ocjenu kvalitete koloriziranih crno-bijelih slika

master's thesis

• Title (english): Objective measures for assessing the quality of colorized black and white images
• Author: Nikolina Bilanović
• Mentor: Sonja Grgić (mentor)
• Committee member: Sonja Grgić (predsjednik povjerenstva)
• Committee member: Tomislav Kos (član povjerenstva)
• Committee member: Marko Bosiljevac (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-07-07, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Kolorizacija je proces dodavanja boje crno-bijelim slikama. Automatizirane metode pretvorbe su izazovno područje koje objedinjuje metode analize i obrade slike te strojno i duboko učenje radi dobivanja vizualno uvjerljivih slika. Procjena kvalitete rezultata kolorizacije provodi se objektivnim mjerama za ocjenu kvalitete slike. U radu je dan pregled metoda kolorizacije, kao i objektivnih mjera za procjenu kvalitete slike. Nad koloriziranim slikama izračunate su objektivne mjere za ocjenu kvalitete i procijenjena je učinkovitost postupaka kolorizacije u postizanju točnosti reprodukcije boja usporedbom rezultata objektivnog i subjektivnog ispitivanja.
Predložena je nova objektivna mjera za objektivnu procjenu kvalitete slike utemeljena na usporedbi vektora boje izvorne i kolorizirane slike nazvana CER - omjer pogreške boje. Objektivna mjera CER pokazala je dobre performanse i rezultate u korelaciji sa subjektivnim doživljajem slika, predlaže se kao prikladna mjera za procjenu kvalitete koloriziranih slika.

Abstract (english)

Colorization is a process of adding color to black and white images. Automatic conversion methods are a challenging field that combines image analysis and processing methods together with machine and deep learning to obtain visually persuasive images. Quality assessment of colorization results is carried out using objective image quality assessment methods. This paper gives an overview of colorization methods, as well as objective measures for image quality assessment. Objective measures for quality assessment were calculated over the colorized images and the efficiency of colorization procedures in achieving the accuracy of color reproduction was evaluated by comparing the results of objective and subjective testing.
A new objective measure for the objective assessment of image quality based on the comparison of the color vector of the original and colored image called CER - color error ratio is proposed. Objective measure CER has shown good performance and its results are in correlation with the subjective experience of images; therefore, it is proposed as a suitable measure for assessing the quality of colored images.

• Keywords: automatizirane metode; kolorizacija; objektivne metode; procjena kvalitete slike; prostor boja; subjektivno ispitivanje; vektor boje
• Keywords (english): automatic methods; colorization; color space; color vector; image quality assessment; objective methods; subjective testing
• Language: croatian
• URN:NBN: urn:nbn:hr:168:938933
• 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: 2023-01-13 14:34:06