| Popis: |
Screening based on non-contact infrared thermometers (NCITs) and IRTGs shows promising results for mass fever screening. IRTGs were found to be powerful, quick and non-invasive methods to detect elevated temperatures. In the case of temperature measurement using IRTGs, regions medically adjacent to inner canthi are preferred sites for fever screening (IEC 80601-2-59:2008), which show good stability and correlation with body temperature. However, detection of canthi in thermal images is challenging due to the absence of features unlike visible images, which have sharp features that can be used for eye corner detection. We use registration of thermal images with visible light images (also called white-light images) to localize canthi regions in thermal images. We study the accuracy of such multi-modal image registration in the context of canthi detection and measure the feasibility of automatic canthi based temperature measurement as an alternative to manual measurement. The second part of thesis refers to the study of image photoplethysmography (IPPG), a cost-effective and flexible method for heart rate monitoring using videos recorded in ambient light. We use low-cost commercial grade video recording equipment (mobile camera/Digital camera) with an ambient light source. The study includes information about signal processing algorithms for estimating heart rate, relevant parameters, and comparison with standard techniques. Such low-cost, multi-purpose solutions for quick screening of subjects provides us with sensible and useful information on elevated body temperature and heart rate. Hence, these methods show promising results that enable mass fever screening as a possibility through temperature and heart rate monitoring, where low cost of installation and flexibility are important. |
