Fast Volume Alignment by Frequency-Marched Newton
Fabian Kruse, Valentin Debarnot, Vinith Kishore, Ivan Dokmanić · 2026
Read Fast Volume Alignment by Frequency-Marched Newton by Fabian Kruse, Valentin Debarnot, Vinith Kishore, Ivan Dokmanić (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-15285.txt.
Description
Kruse and co-authors present Matcha, a fast method for aligning three-dimensional volumes by pose. Classic matched filtering searches a huge grid of rotations and shifts. Matcha instead treats rotation as continuous optimization on SO(3), using a bandlimited Wigner-D expansion for fast values, gradients, and Hessians. A coarse low-bandwidth search seeds candidates, which Newton steps refine as bandwidth grows. On synthetic tests it reached sub-degree accuracy much faster than exhaustive search. Inside a RELION-5 workflow, pose refinement ran over ten times faster at equal quality.
Catalog record
| CREATOR | Fabian Kruse, Valentin Debarnot, Vinith Kishore, Ivan Dokmanić |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 75332 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.15285