基于卡尔曼滤波的ECoG信号去噪方法研究

脑电信号反映了生物体的大脑活动,在采集和处理过程中极易受到各种噪声的干扰,如眨眼、快速眼动、心电、肌电等,这些噪声给脑电信号的分析处理带来了很大的困难。本文提出了卡尔曼滤波模型和模型参数估计的方法,将其应用于脑电ECoG信号去噪预处理。实验所用的数据是公开的脑机接口竞赛实验数据（BCI Competition Ⅲ dataset Ⅰ）,分类正确率为92%。实验结果表明通过本方法去噪预处理后,分类正确率比竞赛第一名高,并且优于小波去噪与谱减法等预处理方法。

Research on ECoG de-noising method based on Kalman filter

The ECoG signals reflect brain activity of organisms.Several noise are interfused into the signals during the recording process,such as eyeblink,eyemovements,cardiac electric activity and myoelectric activity,which brings great difficulties to analysis and process the ECoG signal.In this paper,a new method of kalman filter modeling and model parameter estimation is put forward.And it is applied to the de-noising preprocessing of ECoG.The open brain Computer Interface Competition experimental data（BCI Competition III dataset I） is used in the experiment.Classification accuracy is 92%.The experimental results show that classification accuracy of the method is higher than the first of the competition,and better than the de-noising method of wavelet or spectral subtraction.