diplomsko delo
Jadran Kotnik (Author), Božidar Potočnik (Mentor)

Abstract

V tem diplomskem delu se ukvarjamo z avtomatskim razpoznavanjem nosu in njegovo segmentacijo v digitalnih posnetkih. Rezultat našega dela je algoritem, ki vrača natančen obris nosu. Za razpoznavanje obrobe nosu smo uporabili modele, za iskanje konice nosu in nosnic pa smo uporabili iskanje svetlejših oziroma temnejših področij. Uspešnost našega algoritma smo nato preverili na zbirki 50 slik. Ugotovili smo, da je razpoznavanje obrobe nosu v večini slik dobro, razen v izjemnih primerih, kjer razpoznamo napačni del slike. Če uporabimo za prepoznavanje nosu modele, potem je bila Hausdorffova razdalja v povprečju enaka 3,221 mm s standardnim odklonom 2,320 mm, t.i. povprečna Hausdorffova razdalja pa je bila v povprečju 1,080 mm s standardnim odklonom 0,696 mm. Algoritem na izhodu oblikuje maske prepoznanih komponent nosu, katere lahko uporabimo v naprednejših aplikacijah.

Keywords

računalniški vid;prepoznavanje nosu;razpoznavanje vzorcev;digitalna obdelava slik;

Data

Language: Slovenian
Year of publishing:
Typology: 2.11 - Undergraduate Thesis
Organization: UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher: J. Kotnik
UDC: 004.932(043.2)
COBISS: 19311382 Link will open in a new window
Views: 1444
Downloads: 85
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: AUTOMATED NOSE RECOGNITION FROM DIGITAL IMAGES BY COMPUTER VISION PROCEDURES
Secondary abstract: In this thesis, we are dealing with automatic nose recognition and segmentation in digital images. The result of our work is an algorithm that returns an accurate nose trim. We used models for nose trim recognition and searching of brighter or darker areas for the nose tip and nostril recognition. We verified the success rate of our algorithm on a set of 50 images. We have found that for most images the nose trim is recognized in a good measure, with the exception of extreme cases, where we recognize the wrong part of the image. When we use models for nose recognition the Hausdorff distance averages at 3.221 mm with a standard deviation of 2.320 mm while the so called average Hausdorff distance averages at 1.080 mm with a standard deviation of 0.696 mm. The algorithm forms mask images as the output. The output of our algorithm are mask images that can be used in more advanced applications.
Secondary keywords: computer vision;nose recognition;digital image processing;
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: IX, 27 f.
ID: 8888825