Emotion classification using EEG signals based on tunable-Q wavelet transform
Emotion classification using EEG signals based on tunable-Q wavelet transform
复制标题
DOI：
10.1049/iet-smt.2018.5237
复制
发表时间：
2019-05-01
影响因子：
1.4
通讯作者：
Bajaj, Varun
中科院分区：
文献类型：
--
作者：
Krishna, Anala Hari;Sri, Aravapalli Bhavya;Bajaj, Varun
情感是人类最本能的感觉。情绪分类在脑机接口系统中得到应用，以帮助残疾人。为了识别情绪状态，脑电图（EEG）信号起着至关重要的作用，因为它对人类大脑中的每一种变化状态都提供了即时响应。在这里，可调Q小波变换（TQWT）的实用性进行了探索不同的情绪脑电信号的分类。TQWT将脑电信号分解为子带，提取子带的时域特征。提取的特征被用作极端学习机器分类器的输入，用于快乐、恐惧、悲伤和放松情绪的分类。实验结果表明，该方法具有更好的四种情感的分类性能时，与其他现有的方法相比。
Emotion is a most instinctive feeling of a human. Emotion classification finds application in brain-computer interface systems for the assistance of disabled persons. To recognise the emotional state, electroencephalogram (EEG) signal plays a vital role because it provides immediate response to every state of change in the human brain. Here, the utility of tunable-Q wavelet transform (TQWT) is explored for the classification of different emotions EEG signals. TQWT decomposes EEG signal into subbands and time-domain features are extracted from subbands. The extracted features are used as an input to extreme learning machine classifier for the classification of happy, fear, sad, and relax emotions. Experimental results of the proposed method show better four emotions classification performance when compared with the other existing methods.