工件(错误)
独立成分分析
计算机科学
人工智能
预处理器
峰度
模式识别(心理学)
脑电图
噪音(视频)
离群值
信号(编程语言)
计算机视觉
信号处理
语音识别
图像(数学)
数字信号处理
数学
统计
计算机硬件
精神科
心理学
程序设计语言
作者
Xinxin Wang,Xiuwei Wang,Biao Zhou
标识
DOI:10.1109/wcsp55476.2022.10039236
摘要
At present, the head-mounted human-computer interaction technology that establishes direct communication between brain and computer through EEG acquisition equipment is considered as a cutting-edge research topic. Edge calculation can process data immediately after data collection. When the EEG signals are disturbed by artifacts, it can be preprocessed at the edge device. In this paper, preprocessing steps such as independent component analysis (ICA), kurtosis value identification and improved boxplot technology are proposed to identify and remove artifact points in EEG signal. The experiment uses real datasets, and the artifact removal performance is objectively evaluated by calculating signal-to-noise ratio and correlation coefficient. The processed motor imagery EEG signals, compared with the classification results without artifact removal, have sufficient improvement in multiple groups of binary classification, which indicates the effectiveness of the proposed artifact removal scheme.
科研通智能强力驱动
Strongly Powered by AbleSci AI