The thermal behavior of wood is a critical factor in advanced material assembly. However, pixel-level thermal analysis remains fundamentally constrained by the low resolution and noise inherent to infrared thermography. To address this, we introduce an end-to-end computational framework that synthesizes high-resolution thermal responses directly from standard RGB images. We first establish a core physical linkage: because spatial color variation in natural wood is driven by cellular anatomy, optical intensity serves as a reliable geometric proxy for the localized solid volume fraction. Through leveraging this theoretical insight, we develop an automated finite-element-method data engine that maps pixel-level optical intensity to a 3D thermodynamic voxel grid, generating high-fidelity synthetic thermal responses. We find that 1) when the thermal conductivity along the thickness direction is uniform or linear, wood RGB images and their corresponding thermal responses exhibit extreme morphological similarities, and the lateral diffusion acts as a low-pass filter that smooths out high-frequency details; 2) when the thermal conductivity along the thickness direction is random, such morphological similarities are destroyed, and wood's 3D structure dominantly governs its thermal response. We further utilize these synthetic thermal responses to supervise a neural surrogate model built upon the DINOv3 foundation model. Our results demonstrate that the neural surrogate model successfully internalizes the governing thermodynamic laws, thereby bypassing computationally expensive simulations and enabling high-resolution thermal inference. This methodology effectively bridges the semantic and thermodynamic domains, unlocking systematic, pixel-level analysis of fine-grained wood thermal performance.
The thermal behavior of wood plays a crucial role in advancing wood material assembly strategies , , , , , . While these strategies hold significant potential, they currently remain largely conceptual. Furthermore, pixel-level analysis of wood thermal behavior remains fundamentally unexplored. Although experimental setups utilizing infrared cameras can readily capture low-resolution, noisy thermal data, it is exceptionally challenging to obtain high-fidelity, high-resolution thermal measurements. The absence of such high-quality data impedes a systematic, fine-grained analysis of wood thermal responses, and foundational knowledge on wood thermal behavior remains lacking.
This study is motivated by the empirical observation that, visually, wood RGB images and their corresponding thermal responses exhibit discernible morphological similarities and an inverse relationship between optical intensity and temperature difference . This morphological alignment suggests a fundamental physical linkage between a wood sample's optical intensity and its localized thermodynamic behavior. Because spatial color variation in uncoated, natural wood is primarily governed by cellular anatomy , optical intensity can serve as a geometric proxy for the localized solid volume fraction.
Translating these complex, high-frequency optical patterns into continuous thermal responses, however, presents a significant methodological challenge. A seemingly intuitive solution is to use data-driven computer vision models (such as Generative Adversarial Networks , Variational Autoencoders , diffusion models , , and flow models , ) to synthesize high-resolution thermal responses. However, these models are physically agnostic and fully supervised. To bridge this gap, this paper proposes a novel, end-to-end computational framework to synthesize high-resolution thermal responses directly from standard RGB images.
Parametric ablation study of thermal diffusion. The top and middle rows display the original optical RGB image and the corresponding simulated steady-state surface temperature maps under various diffusion coefficients (α, β).
Wood internal heat transfer profiles mapping the temperature gradients from the bottom thermal boundary (50°C) to the top surface. The uniform and linear kz distributions (left and center columns) facilitate relatively continuous vertical heat conduction. In contrast, the randomly stratified distributions (right column) disrupt direct pathways, introducing tortuous thermal flow and non-linear temperature drops across the wood thickness.
We leverage this synthesized paired dataset to supervise a DINOv3-based encoder-decoder architecture, enabling direct prediction of the thermal responses from the RGB images.
@misc{xie2026synthesizing,
title={Seeing the Heat: Synthesizing High-Resolution Wood Thermal Responses from Optical Imagery},
author={Jingren Xie},
year={2026},
url={https://zekifayes.github.io/seeheat}
}