Representing Flow Fields with Divergence-Free Kernels for Reconstruction

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Ni, Xingyu, Xing, Jingrui, Li, Xingqiao, Wang, Bin, Chen, Baoquan
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908297317056512
author Ni, Xingyu
Xing, Jingrui
Li, Xingqiao
Wang, Bin
Chen, Baoquan
author_facet Ni, Xingyu
Xing, Jingrui
Li, Xingqiao
Wang, Bin
Chen, Baoquan
contents Accurately reconstructing continuous flow fields from sparse or indirect measurements remains an open challenge, as existing techniques often suffer from oversmoothing artifacts, reliance on heterogeneous architectures, and the computational burden of enforcing physics-informed losses in implicit neural representations (INRs). In this paper, we introduce a novel flow field reconstruction framework based on divergence-free kernels (DFKs), which inherently enforce incompressibility while capturing fine structures without relying on hierarchical or heterogeneous representations. Through qualitative analysis and quantitative ablation studies, we identify the matrix-valued radial basis functions derived from Wendland's $\mathcal{C}^4$ polynomial (DFKs-Wen4) as the optimal form of analytically divergence-free approximation for velocity fields, owing to their favorable numerical properties, including compact support, positive definiteness, and second-order differentiablility. Experiments across various reconstruction tasks, spanning data compression, inpainting, super-resolution, and time-continuous flow inference, has demonstrated that DFKs-Wen4 outperform INRs and other divergence-free representations in both reconstruction accuracy and computational efficiency while requiring the fewest trainable parameters.
format Preprint
id arxiv_https___arxiv_org_abs_2504_01913
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Representing Flow Fields with Divergence-Free Kernels for Reconstruction
Ni, Xingyu
Xing, Jingrui
Li, Xingqiao
Wang, Bin
Chen, Baoquan
Graphics
Machine Learning
Accurately reconstructing continuous flow fields from sparse or indirect measurements remains an open challenge, as existing techniques often suffer from oversmoothing artifacts, reliance on heterogeneous architectures, and the computational burden of enforcing physics-informed losses in implicit neural representations (INRs). In this paper, we introduce a novel flow field reconstruction framework based on divergence-free kernels (DFKs), which inherently enforce incompressibility while capturing fine structures without relying on hierarchical or heterogeneous representations. Through qualitative analysis and quantitative ablation studies, we identify the matrix-valued radial basis functions derived from Wendland's $\mathcal{C}^4$ polynomial (DFKs-Wen4) as the optimal form of analytically divergence-free approximation for velocity fields, owing to their favorable numerical properties, including compact support, positive definiteness, and second-order differentiablility. Experiments across various reconstruction tasks, spanning data compression, inpainting, super-resolution, and time-continuous flow inference, has demonstrated that DFKs-Wen4 outperform INRs and other divergence-free representations in both reconstruction accuracy and computational efficiency while requiring the fewest trainable parameters.
title Representing Flow Fields with Divergence-Free Kernels for Reconstruction
topic Graphics
Machine Learning
url https://arxiv.org/abs/2504.01913