diplomsko delo
Žiga Leber (Author), Božidar Potočnik (Mentor)

Abstract

Diplomsko delo se osredotoča na avtomatsko prepoznavanje ust iz digitalnih posnetkov. Pogledali smo obstoječe metode in na podlagi detektorja Viola-Jones ter aktivnih modelov oblike razvili lasten algoritem. Implementiran je bil v jeziku C++ s pomočjo knjižnice OpenCV. Delovanje je bilo preverjeno na ročno označenih slikah. Rezultati so pokazali, da algoritem v nadzorovanem okolju daje spodbudne rezultate. Dobro delovanje smo opazil, ko je oblika ust dovolj splošna in ima oster rob. Slabši rezultati so bili, ko so usta v kotih zelo tanka. Sklepamo lahko, da bi algoritem lahko bil uporabljen v praktični aplikaciji.

Keywords

računalniški vid;prepoznavanje ust;aktivni modeli oblike;digitalni posnetki;

Data

Language: Slovenian
Year of publishing:
Typology: 2.11 - Undergraduate Thesis
Organization: UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher: Ž. Leber
UDC: 004.93'1(043.2)
COBISS: 19153430 Link will open in a new window
Views: 1188
Downloads: 87
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 IN DIGITAL IMAGES
Secondary abstract: This thesis is focused on automatic mouth detection in digital images. Existing methods were reviewed and a new algorithm was devised based on the Viola-Jones detector and active shape models. It was implemented with the help of the OpenCV library in the C++ programming language. Performance was tested on manually annotated images. Results have shown that the algorithm gives encouraging results in a controlled environment. Good accuracy has been observed with general mouth shapes with sharp borders. Conversely, images of faces with thin lips at the corners have produced lesser results. It can be concluded that the algorithm has potential to be used in an practical application.
Secondary keywords: computer vision;mouth detection;active shape models;digital images;
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: XIV, 48 str.
ID: 8888477