Title Shift-map Image Registration
Authors Linus Svärm, Petter Strandmark
Full-text Available as PDF
Year 2010
Document type Conference paper
Conference name International Conference on Pattern Recognition (ICPR) 2010
Conference Date 2010-08-23/2010-08-26
Conference Location Istanbul, Turkey
Status Inpress
Quality controlled Yes
Language eng
Abstract English Shift-map image processing is a new framework<br> based on energy minimization over a large space of<br> labels. The optimization utilizes -expansion moves<br> and iterative refinement over a Gaussian pyramid. In<br> this paper we extend the range of applications to image<br> registration. To do this, new data and smoothness<br> terms have to be constructed. We note a great improvement<br> when we measure pixel similarities with the dense<br> DAISY descriptor. The main contributions of this paper<br> are:<br> • The extension of the shift-map framework to include<br> image registration. We register images for<br> which SIFT only provides 3 correct matches.<br> • A publicly available implementation of shift-map<br> image processing (e.g. inpainting, registration).<br> We conclude by comparing shift-map registration to a<br> recent method for optical flow with favorable results.

Questions: webmaster
Last update: 2013-04-11

Centre for Mathematical Sciences, Box 118, SE-22100, Lund. Telefon: +46 46-222 00 00 (vx)