magistrsko delo
Matej Pinter (Author), Božidar Potočnik (Mentor)

Abstract

V magistrskem delu se ukvarjamo s problemom izločanja območja dokumenta iz digitalnih posnetkov zajetih v nekontroliranem okolju. Rezultat dela sta namizna ter mobilna aplikacija, ki omogočata določitev in izločitev območja dokumenta z različnimi metodami. Za rešitev problema določitve implementiramo tri različne metode računalniškega vida, in sicer določitev z barvnim filtriranjem, določitev z izračunom povprečne razlike med svetlostjo regij in določitev s pomočjo Houghove transformacije. Delovanje implementiranih metod preverimo na lastni bazi digitalnih posnetkov, vzpostavljeni z digitalizacijo več dokumentov v nekontroliranih razmerah z različnimi mobilnimi napravami. Testna baza vključuje posnetke z izrazito kot tudi neizrazito mejo HDB. V rezultatih testiranja se kljub relativno majhni testni bazi v primeru metode z barvnim filtriranjem in metode, bazirane na Houghovi transformaciji pokaže vzorec manj uspešne določitve območja na posnetkih, kjer je dokument obdan z neizrazito mejo HDB. Bolje se obnese metoda, ki opravlja določitev območja na osnovi izračuna povprečne razlike med svetlostjo regij, saj se je na testnih posnetkih z neizrazito mejo HDB obnesla celo bolje kot na posnetkih z izrazito mejo HDB. V delu dodatno opredelimo vpliv prostorske ločljivosti posnetkov in vpliv uporabe bliskavice ob zajemu na uspešnost določitve območja dokumenta posamezne metode. V implementacijo poleg pristopov za določanje in izločanje vključimo najpogostejše operacije, ki omogočajo manipuliranje z izločenim območjem dokumenta.

Keywords

analiza digitaliziranih dokumentov;nenadzorovano okolje;odprava striga;Houghova transformacija;projektivna preslikava;

Data

Language: Slovenian
Year of publishing:
Typology: 2.09 - Master's Thesis
Organization: UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher: [M. Pinter]
UDC: 004.62:004.932(043.2)
COBISS: 17991190 Link will open in a new window
Views: 1252
Downloads: 137
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: DOCUMENT REGION EXTRACTION FROM DIGITAL IMAGES, TAKEN IN UNCONTROLLED ENVIRONMENT
Secondary abstract: In this thesis, we are dealing with the problem of document region extraction from digital images, which might have been taken in uncontrolled environment. Thesis’s results are presented via implementation of two applications, one for mobile and one for desktop platform. Both applications enable detection and extraction of document region with several methods. To solve the problem, we implemented three different computer vision methods, namely the detection using colour filtering, detection based on calculating the average difference between the brightness of the regions and detection based on Hough transform. Validation of implemented methods has been done with establishing an own document image database, which includes digital images of several documents, taken in various uncontrolled conditions with a variety of mobile devices. The database includes document images with intensive and non-intensive document border. Despite relatively small database, the test results have shown that colour filtering and Hough transform methods are less reliable than the detection based on average brightness difference calculation, especially when comes to document images with non-intensive document border. The thesis further defines the impact of image spatial resolution and the impact of flash on the reliability for each implemented method. Additionally the most common methods for manipulation and improvement of extracted document region are presented.
Secondary keywords: document image analysis;uncontrolled environment;image deskew;Hough transform;projective mapping;
URN: URN:SI:UM:
Type (COBISS): Master's thesis/paper
Thesis comment: Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko
Pages: X, 70 f.
ID: 8729330