Thermal Imaging for Contactless Cardiorespiratory and Sudomotor Response Monitoring

2026-02-16 · Source: cs.CV updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Intermediate, extended

Summary

A study characterizes the contactless extraction of electrodermal activity (EDA), heart rate (HR), and breathing rate (BR) from facial thermal video. Researchers developed a signal-processing pipeline that tracks anatomical regions, applies spatial aggregation, and separates slow sudomotor trends from faster cardiorespiratory components. For HR, an orthogonal matrix image transformation (OMIT) decomposition is applied across multiple facial regions of interest (ROIs), while BR estimation averages nasal and cheek signals before spectral peak detection. The pipeline was evaluated on 31 sessions from the public SIMULATOR STUDY 1 (SIM1) driver monitoring dataset. The best fixed EDA configuration (nose region, exponential moving average) achieved a mean absolute correlation of $0.40\pm 0.23$ against palm EDA, with individual sessions reaching 0.89. BR estimation yielded a mean absolute error of $3.1\pm 1.1$ bpm, but HR estimation resulted in $13.8\pm 7.5$ bpm MAE, primarily due to the low camera frame rate of 7.5 Hz.

Key takeaway

For Computer Vision Engineers developing contactless physiological monitoring systems, you should prioritize higher frame rate cameras for reliable heart rate estimation, as 7.5 Hz is insufficient. When implementing EDA monitoring, account for signal polarity inversions and consider dynamic ROI selection or multi-ROI fusion to improve accuracy across diverse subjects and conditions, rather than relying on a single fixed facial region.

Key insights

Thermal imaging can non-invasively monitor cardiorespiratory and sudomotor responses, with varying accuracy based on signal type and camera frame rate.

Principles

• Different facial ROIs capture distinct autonomic components.
• Thermal-EDA polarity can alternate across sessions.
• Low frame rates limit thermal HR accuracy.

Method

A four-stage pipeline detects facial landmarks, aggregates spatial temperature, decomposes signals into sudomotor, cardiac, and respiratory bands, and validates against contact ground truth.

In practice

• Use nose/cheeks with time-domain smoothers for EDA.
• Implement polarity detection for thermal-EDA systems.
• Consider adaptive ROI selection for improved agreement.

Topics

• Thermal Imaging
• Contactless Physiological Monitoring
• Electrodermal Activity
• Cardiorespiratory Monitoring
• Biosignal Processing

Best for: Computer Vision Engineer, AI Researcher, AI Scientist, Research Scientist

Related on AIssential

• PULSE: Privileged Knowledge Transfer from Rich to Deployable Sensors for Embodied Multi-Sensory Learning — PULSE transfers privileged EDA knowledge to low-cost sensors, improving stress detection without requiring EDA at inference.
• Your phone can use tiny skin-colour changes to measure your heart rate — Smartphone cameras can measure heart rate by detecting subtle skin color changes.
• FHRFormer: A Self-Supervised Masked Transformer Framework for Fetal Heart Rate Time-Series Inpainting and Forecasting — A masked transformer autoencoder can robustly reconstruct missing fetal heart rate data for improved AI analysis.
• PhysFlow: Frequency Decoupled with Dual-Field Rectified Flow for Remote Photoplethysmography — PhysFlow improves rPPG robustness by frequency-decoupling signal components and modeling them separately with rectified flow.
• Deep learning-based detection of cessation of breathing in pre-term infants — Deep learning reliably detects infant apnoea using routine NICU signals, with signal modality being key.
• State-Specific Respiratory Signatures for Affective and Stress Recognition: Interpretable Respiratory Markers, Autocorrelation Lags, and Compact CNN Models — Respiratory activity offers distinct, interpretable signatures for stress and affective state recognition, combining CNNs and handcrafted features.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.CV updates on arXiv.org.