Infrared imaging based hyperventilation monitoring through respiration rate estimation

• Infrared thermography based hyperventilation detection has been considered.
• Minimum eigenvalue features are taken as tracking features.
• Motion tracking has been done using Kanade–Lucas–Tomasi algorithm.
• Respiration rate estimation using spirometer test has been done for validation.
• Qualitative comparative analysis between previous literature and proposed method.

A change in the skin temperature is used as an indicator of physical illness which can be detected through infrared thermography. Thermograms or thermal images can be used as an effective diagnostic tool for monitoring and diagnosis of various diseases. This paper describes an infrared thermography based approach for detecting hyperventilation caused due to stress and anxiety in human beings by computing their respiration rates. The work employs computer vision techniques for tracking the region of interest from thermal video to compute the breath rate. Experiments have been performed on 30 subjects. Corner feature extraction using Minimum Eigenvalue (Shi–Tomasi) algorithm and registration using Kanade Lucas–Tomasi algorithm has been used here. Thermal signature around the extracted region is detected and subsequently filtered through a band pass filter to compute the respiration profile of an individual. If the respiration profile shows unusual pattern and exceeds the threshold we conclude that the person is stressed and tending to hyperventilate. Results obtained are compared with standard contact based methods which have shown significant correlations. It is envisaged that the thermal image based approach not only will help in detecting hyperventilation but can assist in regular stress monitoring as it is non-invasive method.