Thermal is Always Wild: Characterizing and Addressing Challenges in Thermal-Only Novel View Synthesis

2026-03-20

AI Summary

Thermal imaging offers superior visibility in dark and adverse conditions, yet its low dynamic range and inter-frame photometric instability often lead to floating artifacts in novel view synthesis. This work presents the first systematic analysis of the core challenges inherent in using pure thermal imagery for novel view synthesis and introduces a general-purpose solution that requires neither RGB guidance nor dataset-specific hyperparameter tuning. By integrating lightweight dynamic range expansion and photometric stability preprocessing with 3D Gaussian splatting rendering, the proposed method significantly enhances image quality and temporal consistency. It achieves state-of-the-art performance across multiple pure thermal imaging benchmarks, substantially outperforming existing approaches.

Technology Category

• Computer Vision: Computational Photography, Image & Video Synthesis
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Multi-instance/Multi-view Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Thermal cameras provide reliable visibility in darkness and adverse conditions, but thermal imagery remains significantly harder to use for novel view synthesis (NVS) than visible-light images. This difficulty stems primarily from two characteristics of affordable thermal sensors. First, thermal images have extremely low dynamic range, which weakens appearance cues and limits the gradients available for optimization. Second, thermal data exhibit rapid frame-to-frame photometric fluctuations together with slow radiometric drift, both of which destabilize correspondence estimation and create high-frequency floater artifacts during view synthesis, particularly when no RGB guidance (beyond camera pose) is available. Guided by these observations, we introduce a lightweight preprocessing and splatting pipeline that expands usable dynamic range and stabilizes per-frame photometry. Our approach achieves state-of-the-art performance across thermal-only NVS benchmarks, without requiring any dataset-specific tuning.

Problem

• thermal imaging
• novel view synthesis
• dynamic range
• photometric instability
• correspondence estimation

Innovation

• thermal-only novel view synthesis
• low dynamic range
• photometric stabilization
• splatting pipeline
• floater artifacts