Comparison of Fourier and normal angle descriptors for the content-based image retrieval

Appropriate methods are necessary to compare and search images automatically without using tags. In this work the retrieval of similar objects in images using the shape information is investigated. For this purpose, Fourier descriptors and the angles from the normal vectors along the boundaries of t...

Full description

Saved in:
Bibliographic Details
Published inTechnisches Messen Vol. 83; no. 4; pp. 225 - 233
Main Authors Hernández Mesa, Pilar, Puente León, Fernando
Format Journal Article
LanguageEnglish
Published De Gruyter Oldenbourg 28.04.2016
Subjects
Content-based image retrieval
correlation for circular data
Formdeskriptoren
Fourier descriptors
Fourier-Deskriptoren
Fourier-Koeffizienten
Inhaltsbasierte Bildsuche
Korrelation für 2
normal angles
Normalenrichtungen
perspektivische Transformation
shape descriptors
warping
zyklischer Größen
Online AccessGet full text

Cover

More Information
Summary:Appropriate methods are necessary to compare and search images automatically without using tags. In this work the retrieval of similar objects in images using the shape information is investigated. For this purpose, Fourier descriptors and the angles from the normal vectors along the boundaries of the objects are used as shape descriptors. Three different features are extracted from the Fourier descriptors to compare the objects. Three different distance functions are proposed to measure the similarity between the objects when the angles of the normal vectors are used as features. Finally, the appropriateness of these methods for the content-based image retrieval is compared at a commonly used database to test such cases.
ISSN:0171-8096
2196-7113
DOI:10.1515/teme-2015-0107