Research Analyzer
← Back ICRA 2026

TI-3DGS: 3D Thermal Reconstruction Via Thermal Imaging-Guided 3D Gaussian Splatting

Yong Tang, Yunhao Li, Xiaodong Wang, Qiang Song, Bing Qin, Xiaocheng Feng, XIN YUAN

PDF

AI summary

Key figure (auto-extracted from paper)
TI-3DGS achieves state-of-the-art 3D thermal reconstruction by modeling radiation attenuation and enhancing edges, overcoming the low-texture and blur challenges of pure thermal imaging.
Thermal reconstruction 3D Gaussian Splatting Thermal imaging Radiation attenuation Edge enhancement Low-texture scenes

Problem

Pure 3D thermal reconstruction fails in low-light or adverse conditions because thermal images lack rich textures and suffer from heat conduction blur, leading to sparse point clouds and severe artifacts in existing methods.

Approach

The authors propose TI-3DGS, a 3D Gaussian Splatting framework that uses a Thermal Imaging Field to model radiation attenuation, a Thermal Attenuation-aware Density Control strategy to densify sparse Gaussians, and an Edge Enhancement Loss to preserve sharp thermal boundaries.

Key results

  • Introduces a Thermal Imaging Field to decouple atmospheric, distance, and sensor attenuation
  • Develops a Thermal Attenuation-aware Density Control strategy for adaptive Gaussian densification
  • Proposes an Edge Enhancement Loss to mitigate blur and preserve thermal boundaries
  • Achieves state-of-the-art performance on the TI-NSD dataset for indoor scenes and comparable results for outdoor scenes

Why it matters

Enables reliable 3D thermal scene reconstruction for critical applications like nighttime surveillance, autonomous driving, and fire rescue where visible light fails.

Abstract

Thermal imaging, with its all-weather capabilities and strong penetration, enables 3D reconstruction in low- light and adverse conditions. In this paper, we investigate RGB-independent pure 3D thermal reconstruction, aiming to overcome the challenges of 3D reconstruction in extreme environments where RGB images are unavailable. However, directly applying visible-light 3D reconstruction methods to thermal images often leads to severe artifacts due to two key challenges: (i) thermal images lack rich textures, hindering detail reconstruction, and (ii) heat conduction causes intensity diffusion, resulting in blurred edges. To address these issues, we propose TI-3DGS, a novel 3D Gaussian Splatting framework guided by thermal imaging. We introduce a Thermal Imaging Field (TIF) to model radiance in thermal domains and a Thermal Attenuation-aware Density Control (TADC) strategy to densify sparse point clouds from low-texture thermal inputs. Additionally, we incorporate an edge-enhancement constraint to mitigate blur from heat diffusion. Extensive experiments on the TI-NSD dataset, covering indoor and outdoor scenarios, show that our TI-3DGS achieves state-of-the-art performance, effectively overcoming texture sparsity and edge degradation in thermal reconstruction.

Index terms

Computer Vision for Automation Computer Vision for Transportation Visual Learning

Related papers