Prepoznavanje lica iz više pogleda korištenjem trajektorija u prostoru svojstvenih slika

scientific master's theses

• Title (english): Multi-view face recognition using trajectories in eigenspace
• Author: Zoran Biuk
• Mentor: Sven Lončarić (mentor)
• Committee member: Slobodan Ribarić (predsjednik povjerenstva)
• Committee member: Sven Lončarić (član povjerenstva)
• Committee member: Robert Manger (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; (Department of Electronic Systems and Information Processing); Zagreb
• Defense date and country: 2002-07-09, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering; Electronics
• Universal decimal classification (UDC): 621.3 - Electrical engineering

Abstract

U ovom radu je prikazana metoda prepoznavanja lica na osnovu više poza. Umjesto prepoznavanja iz samo jednog pogleda, koristi se niz slika koje prikazuju pokrete lica iz jednog profila u drugi. U tom pristupu lica su predstavljena svojim pattern vektorima (projekcije na svojstvena lica) u prostoru svojstvenih slika. Pattern vektori koji predstavljaju niz poza, formiraju trajektoriju u prostoru svojstvenih slika gdje svaka trajektorija odgovara jednoj sekvenci slika (od jednog do drugog profila).
Problem prepoznavanja lica tada postaje problem usporedbe prototipnih i nove trajektorije, gdje se prototipne trajektorije formiraju za vrijeme treniranja sustava, a nova trajektorija odgovara nizu slika osobe koju je potrebno prepoznati.
U radu je prikazan način detekcije lica u slici te novi modeli združivanja (trajektorija) kao i analiza utjecaja određenih parametara na stupanj prepoznavanja. Pored toga, prezentirane su primjene interpolacije trajektorije na prepoznavanje i kompenzaciju problema nedostatka poze.

Abstract (english)

Face recognition based on multi-pose image sequence is presented in this work. Instead of recognizing a face from a single view, a sequence of images showing face movement (from left to the right profile) is used for recognition. In this approach faces are represented by their pattern vectors (projections to eigenfaces) in eigenspace. Pattern vectors corresponding to multiple poses construct a trajectory in eigenspace where each trajectory belongs to one face sequence (from profile to profile).
Face recognition thus become problem of comparison between prototypes and novel trajectories, where prototypes trajectories are formed during training phase while novel trajectory correspond to unknown face sequence to be recognized.
Face detection and new models of matching are presented in this work as well as analysis of some parameters influence to recognition ratio. Furthermore, trajectories interpolation and its applications to recognition and missing pose compensation are presented too.

• Keywords: Prepoznavanje lica; niz slika; velika rotacija u dubinu; svojstvena lica; usporedba trajektorija; interpolacija; PCA
• Keywords (english): Face recognition; image sequence; large rotation in depth; eigenfaces; trajectories matching; interpolation; PCA
• Language: croatian
• URN:NBN: urn:nbn:hr:168:817854
• Study programme: Title: Postgraduate master programme in electrical engineering; Study programme type: university; Study level: postgraduate; Academic / professional title: Magistar znanosti elektrotehnike (Magistar znanosti elektrotehnike)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2019-05-22 13:31:54