Pixel-Based Hyperparameter Selection for Feature-Based Image Registration

Brunet, Florent; Bartoli, Adrien; Navab, Nassir; Malgouyres, Rémy

Pixel-Based Hyperparameter Selection for Feature-Based Image Registration

Files

033-040.pdf (1.54 MB)

Date

2010

Authors

Brunet, Florent
Bartoli, Adrien
Navab, Nassir
Malgouyres, Rémy

Publisher

The Eurographics Association

Abstract

This paper deals with parametric image registration from point correspondences in deformable environments. In this problem, it is essential to determine correct values for hyperparameters such as the number of control points of the warp, a smoothing parameter weighting a term in the cost function, or an M-estimator threshold. This is usually carried out either manually by a trial-and-error procedure or automatically by optimizing a criterion such as the Cross-Validation score. In this paper, we propose a new criterion that makes use of all the available image photometric information. We use the point correspondences as a training set to determine the warp parameters and the photometric information as a test set to tune the hyperparameters. Our approach is fully robust in the sense that it copes with both erroneous point correspondences and outliers in the images caused by, for instance, occlusions or specularities.

        @inproceedings{:10.2312/PE/VMV/VMV10/033-040
,
booktitle = {Vision, Modeling, and Visualization (2010)
},
editor = {Reinhard Koch and Andreas Kolb and Christof Rezk-Salama
},
title = {{Pixel-Based Hyperparameter Selection for Feature-Based Image Registration
}},
author = {Brunet, Florent and 
Bartoli, Adrien and 
Navab, Nassir and 
Malgouyres, Rémy
},
year = {2010
},
publisher = {The Eurographics Association
},
ISBN = {978-3-905673-79-1
},
DOI = {/10.2312/PE/VMV/VMV10/033-040
}
}