Registration by tracking for sequential 2D imaging

Niklas Gunnarsson, Jens Sjölund, Thomas B. Schön

Keywords: Image registration, learning methods, visual tracking

TL;DR: An image registration method for sequential imaging using tracking framework and a sparse-to-dense interpolation scheme.

Abstract: In this paper we present a new modality-agnostic image registration method based on sequential 2D imaging. Our method makes use of a tracking framework in combination with a learned sparse-to-dense interpolation scheme. For tracking we employ a Discriminative Correlation Filter (DCF), a fast method that has lately proven extremely useful for visual tracking. Based on displacement vectors from several trackers the dense displacement field is estimated by a neural network using normalized convolutions.

Paper Type: both

Primary Subject Area: Image Registration

Secondary Subject Area: Transfer Learning and Domain Adaptation

Paper Status: original work, not submitted yet

Source Code Url: https://github.com/ngunnar/tracking_reg

Data Set Url: http://www.cse.yorku.ca/~mridataset/

Registration: I acknowledge that publication of this at MIDL and in the proceedings requires at least one of the authors to register and present the work during the conference.

Authorship: I confirm that I am the author of this work and that it has not been submitted to another publication before.