Front-end Processing of Physiological Signals for the Automated Detection of High-arousal Negative Valence Conditions
Front-end Processing of Physiological Signals for the Automated Detection of High-arousal Negative Valence Conditions
复制标题
用于自动检测高唤醒负价条件的生理信号前端处理
DOI：
10.1109/electronica.2019.8825647
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
T. Ganchev
中科院分区：
文献类型：
--
作者：
Kalin Kalinkov;V. Markova;T. Ganchev
我们提出了一种新的方案，用于前端处理的生理信号，这是专为高唤醒负价条件的自动检测的需要。前端包括三个阶段：ECG和EDA信号的信号预处理，特征提取和特征向量的后处理。不同的配置的前端进行了评估的高唤醒负效价条件的自动检测。评估设置实现了一个通用的实验方案，该方案使用了MAHNOB-HCI数据集。实验结果表明，将所有特征缩放到动态范围[0，1]的后处理方法优于其他后处理方案和原始特征。
We propose a new scheme for front-end processing of physiological signals which is designed for the needs of automated detection of high-arousal negative valence conditions. The front-end incorporates three stages: signal pre-processing of ECG and EDA signals, feature extraction, and postprocessing of feature vectors. Different configurations of the front-end were evaluated for the automated detection of higharousal negative valence conditions. The evaluation setup implements a common experimental protocol, which makes use of the MAHNOB-HCI dataset. The experimental results prove that the post-processing with scaling of all features to the dynamic range [0, 1] is advantageous to other post-processing schemes and the raw features.