Registration and error estimation in correlated multimodal imaging
Résumé
Image registration methods are used in a wide range of applications, in particular in correlated multimodal imaging in life science, yet we often lack an estimate of the associated registration error. In this work we aim to provide such estimates as a quality metric for image registration. Our method relies on multivariate multiple linear regression analysis which provides both image registration itself and registration error estimates. Since linear regression is flexible, models can be extended to integrate constraints such as rigid transformations. This is also known as the orthogonal Procrustes problem. Wepresent the different methods for error estimation used in the correlated multi-modal imaging field, but also the ones used in the registration literature. Finally we provide an implementation of our registration framework as a plugin under Icy software.
Origine | Fichiers produits par l'(les) auteur(s) |
---|
Loading...