Home // ACHI 2024, The Seventeenth International Conference on Advances in Computer-Human Interactions // View article

A Method for Estimating Blood Flow Condition from Skin Tone Information in Real Face Images

Authors:
Miku Shimizu
Naoaki Itakura
Kazuyuki Mito
Kaito Hino
Kohei Okura
Tota Mizuno

Keywords: Real face image; Blood flow; Heat map; Stress.

Abstract:
In this study, we investigate a method to obtain detailed blood flow changes from nasal images. We have previously estimated autonomic nervous activity as an indicator of stress by focusing on the R-B component values (Difference value between R and B component values, the three primary colors of color) of the nasal region in real face images. In fact, changes in R-B component values were found to correspond to large stress fluctuations. However, the effect of noise due to light reflection and shadows could not be taken into account. If the effects of light intensity can be removed, the accuracy of the assessment of autonomic nervous activity can be improved, which would greatly contribute to the development of emotion estimation methods. As a method for this purpose, we propose the introduction of the R+G+B value (Total value of the three primary colors of light: R-component value, G-component value, and B-component value) index, which is black when the R+G+B value is close to 0 and white when the R+G+B value is close to 765. This study will clarify the usefulness of this index by using a heat map that reflects the number of pixels present, with R+G+B values on the x-axis and R-B component values on the y-axis. When this method was applied to an experiment in which changes in blood flow were intentionally induced, characteristic changes in distribution were observed at locations where changes in blood flow occurred. This suggests that by focusing on the locations in the heat map where changes corresponding to changes in blood flow rate are observed, it will be possible to remove the effects of light intensity regardless of differences in people or environments. By applying the proposed method, it is possible to construct a stress estimation system that can be easily used by anyone and contribute to the development of the interface field.

Pages: 85 to 87

Copyright: Copyright (c) IARIA, 2024

Publication date: May 26, 2024

Published in: conference

ISSN: 2308-4138

ISBN: 978-1-68558-163-3

Location: Barcelona, Spain

Dates: from May 26, 2024 to May 30, 2024