Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Thegra: Graph-based SLAM for Thermal Imagery

Anastasiia Kornilova, Ivan Moskalenko, Arabella Gromova, Gonzalo Ferrer, Alexander Menshchikov

TL;DR

This work proposes a sparse monocular graph-based SLAM system for thermal imagery that leverages general-purpose learned features -- the SuperPoint detector and LightGlue matcher, trained on large-scale visible-spectrum data to improve cross-domain generalization.

Abstract

Thermal imaging provides a practical sensing modality for visual SLAM in visually degraded environments such as low illumination, smoke, or adverse weather. However, thermal imagery often exhibits low texture, low contrast, and high noise, complicating feature-based SLAM. In this work, we propose a sparse monocular graph-based SLAM system for thermal imagery that leverages general-purpose learned features -- the SuperPoint detector and LightGlue matcher, trained on large-scale visible-spectrum data to improve cross-domain generalization. To adapt these components to thermal data, we introduce a preprocessing pipeline to enhance input suitability and modify core SLAM modules to handle sparse and outlier-prone feature matches. We further incorporate keypoint confidence scores from SuperPoint into a confidence-weighted factor graph to improve estimation robustness. Evaluations on public thermal datasets demonstrate that the proposed system achieves reliable performance without requiring dataset-specific training or fine-tuning a desired feature detector, given the scarcity of quality thermal data. Code will be made available upon publication.

Thegra: Graph-based SLAM for Thermal Imagery

TL;DR

This work proposes a sparse monocular graph-based SLAM system for thermal imagery that leverages general-purpose learned features -- the SuperPoint detector and LightGlue matcher, trained on large-scale visible-spectrum data to improve cross-domain generalization.

Abstract

Thermal imaging provides a practical sensing modality for visual SLAM in visually degraded environments such as low illumination, smoke, or adverse weather. However, thermal imagery often exhibits low texture, low contrast, and high noise, complicating feature-based SLAM. In this work, we propose a sparse monocular graph-based SLAM system for thermal imagery that leverages general-purpose learned features -- the SuperPoint detector and LightGlue matcher, trained on large-scale visible-spectrum data to improve cross-domain generalization. To adapt these components to thermal data, we introduce a preprocessing pipeline to enhance input suitability and modify core SLAM modules to handle sparse and outlier-prone feature matches. We further incorporate keypoint confidence scores from SuperPoint into a confidence-weighted factor graph to improve estimation robustness. Evaluations on public thermal datasets demonstrate that the proposed system achieves reliable performance without requiring dataset-specific training or fine-tuning a desired feature detector, given the scarcity of quality thermal data. Code will be made available upon publication.
Paper Structure (18 sections, 1 equation, 4 figures, 3 tables)

This paper contains 18 sections, 1 equation, 4 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Thermal Preprocessing for Localization
  4. Graph-Based Methods
  5. Other Methods
  6. Methodology
  7. Keypoint Detection and Matching
  8. Thermal imagery preprocessing
  9. SLAM architecture overview
  10. Initial Pose Estimation
  11. Pose Refinement
  12. Factor Graph Optimization with Weighted Scores
  13. Experiments
  14. Datasets
  15. Evaluation of Thermal Imaging Preprocessing
  16. ...and 3 more sections

Figures (4)

  • Figure 1: Visualization of SuperPoint feature detections. Left: Detections on images subjected to basic preprocessing (normalization only). Right: Detections on images processed using the proposed preprocessing technique, which includes Chambolle denoising, histogram equalization, and median filtering.
  • Figure 2: Comparison of combination of detectors and matchers for a pair of thermal images, both handcrafted and learnable approaches.
  • Figure 3: Comparison of contrast enhancement techniques. Left: results on the dataset from ROVTIO. Right: results on the ViViD++ dataset.
  • Figure 4: Comparison of the Chambolle, Bilateral, and Bandpass filters. Left: results on the dataset from ROVTIO. Right: results on the ViViD++ benchmark.