prikaz prve stranice dokumenta Određivanje vlastitog gibanja kamere primjenom naučene korespondencijske metrike
No public access
master's thesis
Određivanje vlastitog gibanja kamere primjenom naučene korespondencijske metrike
Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing, 2017. urn:nbn:hr:168:928381
Smolčić, Dario
University of Zagreb
Faculty of Electrical Engineering and Computing

Institutional repository: FER Repository

Cite this document

Smolčić, D. (2017). Određivanje vlastitog gibanja kamere primjenom naučene korespondencijske metrike (Master's thesis). Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing. Retrieved from https://urn.nsk.hr/urn:nbn:hr:168:928381

Smolčić, Dario. "Određivanje vlastitog gibanja kamere primjenom naučene korespondencijske metrike." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2017. https://urn.nsk.hr/urn:nbn:hr:168:928381

Smolčić, Dario. "Određivanje vlastitog gibanja kamere primjenom naučene korespondencijske metrike." Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, 2017. https://urn.nsk.hr/urn:nbn:hr:168:928381

Smolčić, D. (2017). 'Određivanje vlastitog gibanja kamere primjenom naučene korespondencijske metrike', Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, accessed 26 November 2024, https://urn.nsk.hr/urn:nbn:hr:168:928381

Smolčić D. Određivanje vlastitog gibanja kamere primjenom naučene korespondencijske metrike [Master's thesis]. Zagreb: University of Zagreb, Faculty of Electrical Engineering and Computing; 2017 [cited 2024 November 26] Available at: https://urn.nsk.hr/urn:nbn:hr:168:928381

D. Smolčić, "Određivanje vlastitog gibanja kamere primjenom naučene korespondencijske metrike", Master's thesis, University of Zagreb, Faculty of Electrical Engineering and Computing, Zagreb, 2017. Available at: https://urn.nsk.hr/urn:nbn:hr:168:928381

Metadata
TitleOdređivanje vlastitog gibanja kamere primjenom naučene korespondencijske metrike
Title (english)Ego-motion estimation with a trained correspondence metric
AuthorDario Smolčić
MentorSiniša Šegvić (mentor)
Committee memberSiniša Šegvić (predsjednik povjerenstva)
Committee memberZoran Kalafatić (član povjerenstva)
Committee memberTomislav Hrkać (član povjerenstva)
GranterUniversity of Zagreb
Faculty of Electrical Engineering and Computing
Zagreb
Defense date and country2017-09-29, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Abstract
Širi kontekst rada jest odred̄ivanje gibanja kamere analizom slijeda stereoskopskih slika. Većina pristupa za rješavanje tog problema temelji se na korespondenciji točkastih značajki u četvorkama slika pribavljenih u susjednim vremenskim trenutcima. U ovom radu se razmatra specifičan slučaj u kojemu se korespondencije ostvaruju ugrađivanjem slikovnih okana u visokodimenzionalni metrički prostor. U okviru rada su proučeni postupci odredđivanja korespondencija, praćenja značajki, te procjene gibanja. Razvijen je napredni sustav estimacije gibanja na temelju libviso biblioteke. U sustav je ugrađena metoda određivanja korespondencija značajki korištenjem naučene korespondencijske metrike. Preciznost postupaka je evaluirana na dva ispitna skupa. Provedena je evaluacija metrike a zatim evaluacija estimacije gibanja. Prikazani su i analizirani dobiveni rezultati te je predložen daljnji smjer razvoja.
Abstract (english)
The broader context of this paper is camera ego-motion estimation through analysis of acquired stereo image sequences. Most approaches to solving this problem are based on feature point correspondences on two image pairs acquired at subsequent time intervals. In this paper, we consider a specific case in which correspondences are calculated by mapping image frames to high dimensional metric space. We studied different methods of feature matching, feature tracking, and ego-motion estimation. We developed an advanced system for ego-motion estimation based on libviso library. In this system, we incorporated new feature matching method based on trained correspondence metric. The precision of these methods is evaluated on two datasets. First, we evaluate trained correspondence metric and then ego-motion estimation is evaluated. Evaluation results are presented and analyzed and future improvements are proposed.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:168:928381
Study programmeTitle: Computing
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
Type of resourceText
File originBorn digital
Access conditionsClosed access
Terms of useLicence In copyright icon
Public note
Created on2019-03-14 00:48:01