Szczegóły publikacji
Opis bibliograficzny
Depth recovery with an area based version of the stereo matching method with scale-space tensor representation of local neighborhoods / Bogusław CYGANEK // W: Computational Science – ICCS 2004 : 4th International Conference : Kraków, Poland, June 6–9, 2004 : proceedings, Pt. 2 / eds. Marian Bubak, Geert Dick van Albada [et al.]. — Berlin [etc.] : Springer-Verlag, 2004. — (Lecture Notes in Computer Science ; ISSN 0302-9743 ; LNCS 3037). — ISBN: 978-3-540-22115-9; e-ISBN: 978-3-540-24687-9. — S. 548–551. — Bibliogr. s. 551, Abstr. — Toż na CD-ROMie
Autor
Dane bibliometryczne
| ID BaDAP | 17525 |
|---|---|
| Data dodania do BaDAP | 2004-10-06 |
| Rok publikacji | 2004 |
| Typ publikacji | materiały konferencyjne (aut.) |
| Otwarty dostęp |  |
| Czasopismo/seria | Lecture Notes in Computer Science |
Abstract
Depth recovery is one of the classical problems of computer vision. Many methods have been developed that address this issue. However, each of them has specific behavior depending on the image contents, acquisition system, computation time, etc. The variety of methods is due to the non uniqueness of the matching process involved in depth recovery from a pair of images of a given scene. The presented method relies on tensor representation of local structures in images. This representation allows for extraction of the local phase, signal intensity and coherence measure of a local neighborhood in different scales. This operator showed to be very useful in detection of locally oriented structures and corners in single digital images - presented at the previous ICCS. In this paper, the tensor operator is used for area-based matching of stereo images. Due to tensor transformation of the input images the more reliable matching was achieved. The method was tested with many different classes of stereo images. In this paper we present and discuss the experimental results and also details of implementation.