CAJ | 학술논문

脑电采集后得到的脑电信号( Electroencephalogram，EEG)中含有噪声信号，为了有效去除噪声并保留有用信息，本文在软阈值去噪的基础上，提出一种改进阈值去除EEG噪声的算法。利用小波变换对EEG信号分解，得到多层的高频系数和低频系数；根据分解层次不同，对小波系数进行自适应的阈值处理；将缩放后的小波系数重构，得到去噪后的EEG信号。以信噪比、均方根误差作为去噪效果的定量指标，将改进算法与硬阈值法、软阈值法、Garrote阈值法进行比较，结果表明，改进阈值法优于其他3种阈值法。
뇌전채집후득도적뇌전신호( Electroencephalogram，EEG)중함유조성신호，위료유효거제조성병보류유용신식，본문재연역치거조적기출상，제출일충개진역치거제EEG조성적산법。이용소파변환대EEG신호분해，득도다층적고빈계수화저빈계수；근거분해층차불동，대소파계수진행자괄응적역치처리；장축방후적소파계수중구，득도거조후적EEG신호。이신조비、균방근오차작위거조효과적정량지표，장개진산법여경역치법、연역치법、Garrote역치법진행비교，결과표명，개진역치법우우기타3충역치법。
In order to eliminate the noise mixed in Electroencephalogram( EEG) and retain useful EEG information,an EEG de￣noising method based on adaptive threshold is proposed,which is improved on the basis of soft threshol￣ding. Firstly,high frequency coefficients and low frequency coefficients of multilayer signals are obtained by wavelet decomposition. Then,detail coefficients is processed by using the adaptive threshold. Finally,the original EEG signal is resumed by reconstructing shrinked detail coefficients. The final results show that the proposed de￣noising algorithm has perspective of higher SNR and lower RMSE compared to soft thresholding,hard thresholding and Gar￣rote thresholding.