Effects of Changes in Subject and Ambient Surface Temperatures on Discriminating between Normal and Falling Postures Using a Thermal Imaging Sensor

Tomoya Miyasaka, Souta Satomura, Riku Hosoya, Mamiko Kamoshida, Keiko Kawashima

Fall detection, Thermal imaging sensor, Discriminant analysis, Temperature Effects

Some thermal imaging monitoring systems for detecting falls of the elderly use temperature data output by a thermal imaging sensor with discriminant analysis techniques to discriminate between normal and falling postures. This study investigates the effects of subject and ambient temperatures on fall detection based on the results of discriminating between normal and falling postures by changing the subject surface temperature (subject temperature) and the ambient surface temperature (ambient temperature). The subjects were two healthy adults, each in normal and falling postures, whose thermal imaging data were acquired at two subject temperatures and three ambient temperatures. Six discriminant formulae were developed from the acquired data to obtain the discriminant results. The results show that the discrimination rate was 100% for the detection within the temperature range of the data from which the discriminant formulae were developed. In contrast, the discrimination rate decreased for the detection outside the temperature range of the data from which the discriminant formulae were developed. The results indicate that developing discriminant formulae using thermal imaging data in the temperature range of the conditions of use can provide a high discrimination rate between normal and falling postures within that temperature range.

https://doi.org/10.31871/IJNTR.10.9.4