diplomsko delo
Aljaž Frančič (Author), Božidar Potočnik (Mentor), Martin Šavc (Co-mentor)

Abstract

V diplomskem delu so raziskane metode računalniškega vida za avtomatsko prepoznavanje ust iz digitalnih posnetkov, pri čemer so v ospredje postavljeni aktivni modeli. V uvodu so natančneje predstavljeni problem, cilji, omejitve in struktura diplomskega dela. Preučene so tri metode, ki temeljijo na aktivnih modelih: aktivni modeli kontur, aktivni modeli oblike in aktivni modeli izgleda. Poudarek je na slednjima in njuna uporaba je v delu tudi demonstrirana. Predstavljen je nov pristop za detekcijo ustnic, imenovan dvofazni aktivni model izgleda. Za aktivni model izgleda je sintetizirana lastna učna in testna množica vzorcev tako za eno-, kot tudi dvofazni model. V zaključnih poglavjih sta podani še kvalitativna in kvantitativna primerjava treh podanih algoritmov, pri čemer je kot ena izmed metrik uporabljena Hausdorffova razdalja. Najboljši rezultati na naši lastni testni množici so doseženi z dvofaznim aktivnim modelom izgleda.

Keywords

razpoznavanje vzorcev;računalniški vid;usta;razpoznavanje ust;metode aktivnih modelov;aktivni modeli kontur;aktivni modeli oblike;aktivni modeli izgleda;OpenCV;diplomske naloge;

Data

Language: Slovenian
Year of publishing:
Typology: 2.11 - Undergraduate Thesis
Organization: UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher: [A. Frančič]
UDC: 004.932ʹ1(043.2)
COBISS: 19089174 Link will open in a new window
Views: 1780
Downloads: 280
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: AUTOMATIC MOUTH DETECTION FROM DIGITAL IMAGES BY USING ACTIVE MODELS
Secondary abstract: Herein, methods of computer vision for automatic mouth detection from digital images are explored, with prominence given to active models. In the introduction, details concerning the tackled problem, goals, limitations and the structure of the diploma thesis, are presented. Three methods based on active models are examined: active contour models, active shape models and active appearance models. The latter two are explored to a greater extent, including demonstration of usage. A novel approach for lip detection is unveiled, dubbed the two-phase active appearance model. Training and validation sets are synthesized for both the one- and the two-phase active appearance model. In the concluding chapters, the qualitative and quantitative comparison between the three presented algorithms is given, using the Hausdorff distance as one of the metrics. The best results using our validation set were achieved by the two-phase active appearance model.
Secondary keywords: pattern recognition;computer vision;mouth detection;active models methods;
URN: URN:SI:UM:
Type (COBISS): Bachelor thesis/paper
Thesis comment: Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Računalništvo in informacijske tehnologije
Pages: XI, 55 str.
ID: 8773369