Attributed graph matching for image-features association using SIFT descriptors

Authors:
Gerard Sanromà;René Alquézar;Francesc Serratosa
Affiliations:
Departament d'Enginyeria Informàtica i Matemàtiques, Universitat Rovira i Virgili, Tarragona, Spain;Institut de Robtica i Informtica Industrial, CSIC, UPC, Barcelona, Spain;Departament d'Enginyeria Informàtica i Matemàtiques, Universitat Rovira i Virgili, Tarragona, Spain
Venue:
SSPR&SPR'10 Proceedings of the 2010 joint IAPR international conference on Structural, syntactic, and statistical pattern recognition
Year:
2010

Citing 6
Cited 0

A Graduated Assignment Algorithm for Graph Matching

IEEE Transactions on Pattern Analysis and Machine Intelligence
Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography

Communications of the ACM
Structural Graph Matching Using the EM Algorithm and Singular Value Decomposition

IEEE Transactions on Pattern Analysis and Machine Intelligence - Graph Algorithms and Computer Vision
Distinctive Image Features from Scale-Invariant Keypoints

International Journal of Computer Vision
A Performance Evaluation of Local Descriptors

IEEE Transactions on Pattern Analysis and Machine Intelligence
A robust Graph Transformation Matching for non-rigid registration

Image and Vision Computing

Quantified Score

Hi-index	0.00

Visualization

Abstract

Image-features matching based on SIFT descriptors is subject to the misplacement of certain matches due to the local nature of the SIFT representations. Some well-known outlier rejectors aim to remove those misplaced matches by imposing geometrical consistency. We present two graph matching approaches (one continuous and one discrete) aimed at the matching of SIFT features in a geometrically consistent way. The two main novelties are that, both local and contextual coherence are imposed during the optimization process and, a model of structural consistency is presented that accounts for the quality rather than the quantity of the surrounding matches. Experimental results show that our methods achieve good results under various types of noise.