Mean of correlation method for optimization of affective states detection in children
At the moment, most of the studies on classification of affective states for children focus on visual observations and physiological cues, where all data collection for measuring physiological signals are contact-based and invasive. With the requirement of having the measuring device attached to t...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English English |
| Published: |
IEEE
2018
|
| Subjects: | |
| Online Access: | http://irep.iium.edu.my/68664/ http://irep.iium.edu.my/68664/ http://irep.iium.edu.my/68664/ http://irep.iium.edu.my/68664/7/68664_Mean%20of%20correlation%20method%20for%20optimization_scopus.pdf http://irep.iium.edu.my/68664/12/68664_Mean%20of%20correlation%20method%20for%20optimization.pdf |
| Summary: | At the moment, most of the studies on classification of affective states for children focus on
visual observations and physiological cues, where all data collection for measuring physiological signals
are contact-based and invasive. With the requirement of having the measuring device attached to the
body approach, distraction of the subject normally masks the true affective states of the subject due to
discomfort. In this paper, a non-invasive, contactless, and less distraction method is proposed to measure the
physiological cues of the subjects using their thermal imprints from frontal face imaging. A thermal image
camera is used to identify basic affective states, where it is a contactless and seamless device with ability to
read the radiated thermal imprint of the subjects’ facial skin temperature. This paper proposes an effective
algorithm of texture analysis based on novel technique using Gray Level Co-occurrence Matrix approach to
be applied so as to identify blood-flow region. The cues from the first order statistics are computed in the
identified blood flow region and concatenated along with second order statistics cues, in order to construct
feature vectors to administer the vital and distinguishable characteristic pattern between affective states in
thermal images. Result from the fine k-NN classifier obtained promises the efficacy of the proposed approach
to be applied in our future work in human–robot interaction for autistic children learning and training. |
|---|