From Krylov Spaces to Power Series: Advancing Large-Scale Bundle Adjustment With and Without Initialization
MCML Authors
Simon Weber
Dr.
* Former Member
Abstract
Simon Weber
Dr.
* Former Member
Abstract
This dissertation advances large-scale bundle adjustment by exploiting its structure. It introduces multidirectional Krylov solvers, proposes inverse expansion methods using power series, and extends variable projection techniques. These contributions improve scalability, efficiency, and robustness, opening new directions beyond classical approaches.
BibTeXKey: Web25a