Objektivna mjera kvalitete slike utemeljena na višerezolucijskoj analizi

doctoral thesis

• Title (english): Objective image quality measure based on multiresolution analysis
• Author: Emil Dumić
• Mentor: Sonja Grgić (mentor)
• Committee member: Tomislav Kos (predsjednik povjerenstva)
• Committee member: Borivoj Modlic (član povjerenstva)
• Committee member: Lidija Mandić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; (Department of Communication and Space Technologies); Zagreb
• Defense date and country: 2011-10-14, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering
• Universal decimal classification (UDC): 621.3 - Electrical engineering

Abstract

Objektivno vrednovanje kvalitete slike ima značajnu ulogu u različitim postupcima obrade slike i videosignala, kao što su kompresija, interpolacija, uklanjanje šuma u slici i slično, pri čemu se uspješnost pojedinog postupka obrade procjenjuje na temelju njegova utjecaja na kvalitetu slike. U radu je predložena nova mjera za objektivnu ocjenu kvalitete slike koja se temelji na višerezolucijskoj analizi slike. Iskorištena su svojstva diskretne wavelet transformacije (DWT, Discrete Wavelet Transform) te usmjerene piramidalne wavelet transformacije (SPWT, Steerable Pyramid Wavelet Transform), koja su slična ljudskom vizualnom sustavu te su pogodna za ocjenu kvalitete slike. Wavelet-dekompozicija računa se na različitim skalama, te se slika dijeli na koeficijente aproksimacije (na najnižoj frekvencijskoj skali), koeficijente frekvencijskih potpojaseva i koeficijente detalja (koeficijenti najviših frekvencija). Pomoću njih se računa mjera kvalitete slike (IQM, Image Quality Measure) koja je definirana kao umnožak srednje vrijednosti kontrasta i strukture između svih apsolutnih vrijednosti potpojaseva originalne i izobličene slike, osim aproksimacijskih koeficijenata i koeficijenata detalja. IQM i još dvanaest relevantnih objektivnih mjera uspoređeno je uz pomoć nove VCL@FER baze slika i šest drugih baza slika te je ispitana korelacija sa subjektivnim ocjenama. Ispitana je i statistička značajnost razlika varijanci za svaku objektivnu mjeru i svaku bazu slika.

Abstract (english)

Objective image quality evaluation plays an important role in other image and video processing techniques, such as compression, interpolation, noise reduction, where each evaluation method is based on image quality estimation. The thesis presents innovative approach to the objective quality evaluation that could be computed using multiresolution image analysis. It is based on DWT subband decomposition or SPWT decomposition, which properties are similar to human visual system (HVS) characteristics facilitating integration of SPWT into image quality evaluation. SPWT decomposition is done with multiresolution analysis of a signal that allows us to decompose a signal into approximation, bandpass and highpass subbands. SPWT coefficients are computed using wavelet filter banks with 2 orientations (vertical and horizontal) in spatial domain. Wavelet coefficients are used to compute new image quality measure (IQM). IQM is defined as a product of a mean contrast and structure between absolute bandpass coefficients of original and degraded image. IQM measure is compared with 12 other existing image quality measures using new image database VCL@FER and 6 other image databases. Correlation between objective and subjective image quality measures is presented, for all image databases. Statistical significance between variances of residuum (between linearized objective and subjective measures) is also presented. Results show that IQM measure yields best correlation results with subjective grades and that it is not dependent on image content and degradation type.

• Keywords: kvaliteta slika; diskretna wavelet transformacija; piramidalna wavelet transformacija; srednja iskustvena vrijednost; baza slika; SSIM
• Keywords (english): image quality; discrete wavelet transform; steerable pyramid wavelet transform; mean opinion score; image database; SSIM
• Language: croatian
• URN:NBN: urn:nbn:hr:168:482684
• Project: Number: 036-0361630-1635; Title: Upravljanje kvalitetom slike u radiodifuziji digitalnog videosignala; Leader: Sonja Grgić; Jurisdiction: Croatia; Funder: MZOS; Funding stream: ZP
• Project: Number: 036-0982560-1643; Title: Inteligentno određivanje značajki slike u sustavima za otkrivanje znanja; Leader: Mislav Grgić; Jurisdiction: Croatia; Funder: MZOS; Funding stream: ZP
• Study programme: Title: Doctoral study programme "Electrical Engineering and Computing"; Study programme type: university; Study level: postgraduate; Academic / professional title: doktor/doktorica znanosti, po-dručje tehničkih znanosti (doktor/doktorica znanosti, po-dručje tehničkih znanosti)
• Catalog URL: http://lib.fer.hr/cgi-bin/koha/opac-detail.pl?biblionumber=36859
• Type of resource: Text
• Extent: 160 str ; 30 cm
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2019-07-02 07:48:47