Diff-Reg: Diffusion Model in Doubly Stochastic Matrix Space for Registration Problem

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F25%3A00380048" target="_blank" >RIV/68407700:21230/25:00380048 - isvavai.cz</a>

Výsledek na webu

<a href="https://doi.org/10.1007/978-3-031-73650-6_10" target="_blank" >https://doi.org/10.1007/978-3-031-73650-6_10</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1007/978-3-031-73650-6_10" target="_blank" >10.1007/978-3-031-73650-6_10</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Diff-Reg: Diffusion Model in Doubly Stochastic Matrix Space for Registration Problem

Popis výsledku v původním jazyce

Establishing reliable correspondences is essential for 3D and 2D-3D registration tasks. Existing methods commonly leverage geometric or semantic point features to generate potential correspondences. However, these features may face challenges such as large deformation, scale inconsistency, and ambiguous matching problems (e.g., symmetry). Additionally, many previous methods, which rely on single-pass prediction, may struggle with local minima in complex scenarios. To mitigate these challenges, we introduce a diffusion matching model for robust correspondence construction. Our approach treats correspondence estimation as a denoising diffusion process within the doubly stochastic matrix space, which gradually denoises (refines) a doubly stochastic matching matrix to the ground-truth one for high-quality correspondence estimation. It involves a forward diffusion process that gradually introduces Gaussian noise into the ground truth matching matrix and a reverse denoising process that iteratively refines the noisy one. In particular, we deploy a lightweight denoising strategy during the inference phase. Specifically, once points/image features are extracted and fixed, we utilize them to conduct multiple-pass denoising predictions in the reverse sampling process. Evaluation of our method on both 3D and 2D3D registration tasks confirms its effectiveness. The code is available at https://github.com/wuqianliang/Diff-Reg.

Název v anglickém jazyce

Diff-Reg: Diffusion Model in Doubly Stochastic Matrix Space for Registration Problem

Popis výsledku anglicky

Establishing reliable correspondences is essential for 3D and 2D-3D registration tasks. Existing methods commonly leverage geometric or semantic point features to generate potential correspondences. However, these features may face challenges such as large deformation, scale inconsistency, and ambiguous matching problems (e.g., symmetry). Additionally, many previous methods, which rely on single-pass prediction, may struggle with local minima in complex scenarios. To mitigate these challenges, we introduce a diffusion matching model for robust correspondence construction. Our approach treats correspondence estimation as a denoising diffusion process within the doubly stochastic matrix space, which gradually denoises (refines) a doubly stochastic matching matrix to the ground-truth one for high-quality correspondence estimation. It involves a forward diffusion process that gradually introduces Gaussian noise into the ground truth matching matrix and a reverse denoising process that iteratively refines the noisy one. In particular, we deploy a lightweight denoising strategy during the inference phase. Specifically, once points/image features are extracted and fixed, we utilize them to conduct multiple-pass denoising predictions in the reverse sampling process. Evaluation of our method on both 3D and 2D3D registration tasks confirms its effectiveness. The code is available at https://github.com/wuqianliang/Diff-Reg.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

<a href="/cs/project/GM22-23183M" target="_blank" >GM22-23183M: Nová generace algoritmů pro řešení problémů geometrie kamer</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2025

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Computer Vision – ECCV 2024, Part LXV

ISBN

978-3-031-73649-0

ISSN

0302-9743

e-ISSN

1611-3349

Počet stran výsledku

19

Strana od-do

160-178

Název nakladatele

Springer, Cham

Místo vydání

—

Místo konání akce

Milano

Datum konání akce

29. 9. 2024

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

001403061600010