Effiziente Schätzung dichter Bewegungsvektorfelder unter Berücksichtigung der Epipolargeometrie zwischen unterschiedlichen Ansichten einer Szene

Abstract

In this book an efficient algorithm for estimating dense motion vector fields between different views of a scene is presented. The approach combines geometric constraints between two views of a static scene with a model to describe the apparent motion in the image plane. The estimation is performed using a maximum-a-posteriori-method yielding reliable results, which is demonstrated using challenging real world image sequences.