Skip to Main content Skip to Navigation
Conference papers

Comparison of affine-invariant local detectors and descriptors

Krystian Mikolajczyk 1 Cordelia Schmid 1, *
* Corresponding author
1 LEAR - Learning and recognition in vision
GRAVIR - IMAG - Laboratoire d'informatique GRAphique, VIsion et Robotique de Grenoble, Inria Grenoble - Rhône-Alpes, CNRS - Centre National de la Recherche Scientifique : FR71
Abstract : In this paper we summarize recent progress on local photometric invariants. The photometric invariants can be used to find correspondences in the presence of significant viewpoint changes. We evaluate the performance of region detectors and descriptors. We compare several methods for detecting affine regions [4, 9, 11, 18, 17]. We evaluate the repeatability of the detected regions, the accuracy of the detectors and the invariance to geometric as well as photometric image transformations. Furthermore, we compare several local descriptors [3, 5, 8, 14, 19]. The local descriptors are evaluated in terms of two properties: robustness and distinctiveness. The evaluation is carried out for different image transformations and scene types. We observe that the ranking of the detectors and descriptors remains the same regardless the scene type or image transformation.
Document type :
Conference papers
Complete list of metadata
Contributor : Thoth Team <>
Submitted on : Monday, December 20, 2010 - 9:09:30 AM
Last modification on : Monday, December 28, 2020 - 3:44:02 PM


  • HAL Id : inria-00548538, version 1




Krystian Mikolajczyk, Cordelia Schmid. Comparison of affine-invariant local detectors and descriptors. 12th European Signal Processing Conference (EUSIPCO '04), Sep 2004, Vienna, Austria. pp.1729--1732. ⟨inria-00548538⟩



Record views