On RGB-TIR Stereo Calibration under Extreme Resolution Asymmetry

2026-05-15

AI Summary

This work addresses the challenging stereo calibration problem arising from the extreme resolution asymmetry between a low-cost, low-resolution thermal infrared (TIR) camera (80×62 pixels) and a high-resolution RGB camera (2028×1520 pixels). To tackle this issue, the authors propose a practical RGB-TIR joint calibration framework that employs an OLED dynamic calibration board to alternately display modality-specific patterns on the same physical plane. A specialized corner detection algorithm, tailored for extremely low-resolution TIR images and based on Hessian saddle-point analysis combined with Mean Shift refinement, is introduced. Geometrically consistent calibration is achieved through baseline-constrained bundle adjustment. The method requires only a single calibration board to simultaneously calibrate both modalities, yielding a reprojection error of 0.382 pixels with an actual baseline of 32.7 mm (nominal 30 mm), and demonstrates cross-modal projection consistency in building thermographic assessments.

Abstract

Accurate geometric calibration of RGB-thermal infrared (TIR) stereo camera systems is essential for multimodal building envelope analysis, yet remains challenging when low-cost thermal sensors with very low spatial resolution are employed. This paper presents a practical stereo calibration framework for an RGB camera (2028 x 1520 px) paired with a TIR camera operating at only 80 x 62 px - a pixel-count ratio of approximately 1:625. An active OLED screen dynamically switches modality-specific patterns (checkerboard for TIR, ChArUco for RGB) on a single physical surface, providing controlled and repeatable thermal contrast. A dedicated corner detection algorithm combining perspective rectification, Hessian saddle-point analysis, and Mean Shift localisation achieves reliable checkerboard detection at 80 x 62 px without per-frame parameter tuning. A baseline-constrained bundle adjustment enforces physically consistent rig geometry under the planar-calibration-object degeneracy, yielding a stereo baseline of 32.7 mm (nominal 30 mm) with an overall reprojection error of 0.382 px. The system is validated on a thermally active building mock-up using constant-depth and per-pixel depth estimation, demonstrating consistent TIR-to-RGB projection suitable for building energy performance assessment.

Problem

• RGB-TIR stereo calibration
• extreme resolution asymmetry
• low-resolution thermal imaging
• geometric calibration
• multimodal sensor fusion

Innovation

• RGB-TIR stereo calibration
• extreme resolution asymmetry
• active OLED calibration pattern
• low-resolution thermal corner detection
• baseline-constrained bundle adjustment