System for efficient reconstruction of images and light fields using compressive sensing.

doctoral thesis

• Title (croatian): Sustav za učinkovitu rekonstrukciju slika i svjetlosnih polja primjenom sažimajućega očitavanja.
• Author: Ivan Ralašić
• Mentor: Damir Seršić (mentor)
• Committee member: Damir Seršić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; (Department of Electronic Systems and Information Processing); Zagreb
• Defense date and country: 2020, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing; Data Processing
• Universal decimal classification (UDC): 004 - Computer science and technology. Computing. Data processing

Abstract

In this thesis, a design of a measurement system for image and light field acquisition comprised of off-the-shelf components (i.e. camera and projector) is proposed. Additionally, a detailed calibration procedure that enables high-quality reconstructions using a compressive sensing (CS) framework is proposed. It minimizes the influence of additive and multiplicative noise in the CS setup. The calibrated system is used for efficient sampling of images and light fields using a sub-Nyquist number of measurements. The proposed system enables high-resolution imaging using a low-resolution imaging sensor. This fact is especially important in the area of light field imaging, where an inherent spatio-angular trade-off exists. The proposed measurement setup and the reconstruction methods enable overcoming the aforementioned trade-off. Finally, efficient methods for reconstruction of CS images using machine learning are proposed. They have the benefits of improved reconstruction speed and quality of the CS images, especially at extremely low measurement rates.

Abstract (croatian)

U ovom radu, predložen je dizajn mjernog sustava za akviziciju slika i svjetlosnih polja sastavljenog od nemodificiranih komponenti (kamere i projektora). Nadalje, predstavljena je metoda umjeravanja koja omogućuje rekonstrukciju slika visoke kvalitete korištenjem sažimajućeg očitavanja. Razvijena metoda umjeravanja smanjuje utjecaj aditivnog i multiplikativnog šuma u sustavu sažimajućeg očitavanja. Kalibrirani sustav korišten je za učinkovito uzorkovanje slika i svjetlosnih polja korištenjem pod-Nyquistovskog broja uzoraka. Predloženi sustav omogućuje visokorezolucijsko snimanje slika i svjetlosnih polja korištenjem niskorezolucijskog slikovnog senzora. Spomenuta činjenica posebno je važna pri snimanju svjetlosnih polja, gdje postoji kompromis između prostorne i kutne rezolucije. Konačno, u radu su predložene efikasne metode za rekonstrukciju sažeto očitanih slika korištenjem strojnog učenja. Razvijene metode imaju brojne prednosti poput bržeg vremena rekonstrukcije te kvalitete sažeto očitanih slika, posebice u slučaju ekstremnog poduzorkovanja.

• Keywords: compressive sensing; compressive imaging; camera-projector system; light field; reconstruction; optimization; sparsity; machine learning; dual imaging; light transport matrix
• Keywords (croatian): sažimajuće očitavanje; sažimajuća fotografija; sustav kamere i projektora; svjetlosno polje; rekonstrukcija; optimizacija; rijetkost; strojno učenje; dualna fotografija; transportna matrica svjetla
• Language: english
• URN:NBN: urn:nbn:hr:168:873639
• Promotion: 2020
• Study programme: Title: Computer Science; Study programme type: university; Study level: postgraduate; Academic / professional title: Doktor znanosti (Doktor znanosti)
• Catalog URL: https://lib.fer.hr:8443/cgi-bin/koha/catalogue/detail.pl?biblionumber=51853&searchid=scs_1600096011439
• Type of resource: Text
• Extent: 175 str.
• File origin: Born digital
• Access conditions: Access restricted to students and staff of home institution
• Created on: 2020-09-14 15:08:55