Correspondence estimation in image pairs
André Redert, Eefje Hendriks, Jan Biemond · IEEE Signal Processing Magazine · 1999
This article provides an overview of current techniques for dense geometric correspondence estimation. We first formally define geometric correspondence and investigate the different types of image pairs. Then, we briefly look at the classic approaches to correspondence estimation, at their feasibility and flaws for simultaneous dense estimation. We focus on the Bayesian approach, which is very well suited for this task, and for which several promising algorithms have previously been developed.