Share

Smart low-cost thermal imaging acquisition towards personal comfort prediction.

Number: 3363

Author(s) : SOLEIMANIJAVID A., KONSTANTZOS I.

Summary

Ensuring occupants' thermal comfort is rapidly becoming an essential objective in building design and operation, as it plays a crucial role in well-being and productivity. Conventional HVAC system controllers operate based on predefined setpoints and schedules, or individuals’ selections. However, as thermal sensation may vary, not only from person to person but also differ over time, the automated operation needs to be informed by real-time sensing. In recent years, advancements in deep learning models have provided an opportunity to deploy vision-based sensors in buildings toward occupant-centric controls (OCC). Vision-based systems are popular devices used to monitor individual thermal sensation and satisfaction due to their capacity to non-intrusively measure skin temperature, a physiological variable that is related to thermal comfort prediction. However, this advantage of remote sensing also leads to reduced accuracy compared to conventional temperature sensors. One of the critical variables responsible for the reduced accuracy is the camera’s distance from the subject, also known as ‘the working distance’. As the requirement of thermal cameras in front of the targets within a fixed distance is not applicable to real operational conditions, this study proposes an experimental framework to perform real-time correction of the thermal camera’s temperature for targets at the distance longer than the thermal camera’s calibration distance. The prediction framework uses a low-cost thermal camera and an RGB-D module to extract the target’s surface temperature and their distance to the camera, and its output can be used to assess an individual's thermal comfort based on skin temperature variation. This approach can be combined with computer vision approaches to allow the continuous detection of the occupants' faces or motion patterns, providing a holistic, multi-modal sensing solution towards occupant-centric controls.

Available documents

Format PDF

Pages: 10 p.

Available

Free

Download

Details

  • Original title: Smart low-cost thermal imaging acquisition towards personal comfort prediction.
  • Record ID : 30030229
  • Languages: English
  • Source: 2022 Purdue Conferences. 7th International High Performance Buildings Conference at Purdue.
  • Publication date: 2022

Links


See other articles from the proceedings (39)
See the conference proceedings

Indexing