ThermalWrist: Smartphone Thermal Camera Correction Using a Wristband Sensor †

Area of Science:

• Biomedical Engineering
• Computer Vision
• Sensor Technology

Background:

• Smartphone thermal cameras are valuable for healthcare but lack accuracy for human body temperature monitoring.
• Small sensor sizes make smartphone thermal images susceptible to temperature fluctuations.

Purpose of the Study:

• To introduce ThermalWrist, a novel dynamic offset correction method for smartphone thermal imaging.
• To enhance the accuracy of absolute temperature measurements from smartphone thermal cameras.

Main Methods:

• Utilized the inherent reliability of relative temperatures in thermal images.
• Combined smartphone thermal images with absolute temperature data from a wristband sensor.
• Developed a dynamic offset correction algorithm (ThermalWrist).

Main Results:

• Reduced mean absolute error in face temperature measurement by 49.4%.
• Decreased standard deviation of face temperature measurement error by 64.9%.
• Increased Pearson's correlation coefficient by 112%, indicating improved accuracy.

Conclusions:

• ThermalWrist significantly improves the accuracy of human body temperature monitoring using smartphone thermal cameras.
• The method is effective in typical indoor environments (≥22.91 °C).
• This technology holds promise for remote health monitoring and research applications.