基于FPGA的脑电信号CSP算法实现

Implementation of CSP for EEG signal separation based on FPGA

脑机接口(BCI)是将采集到的大脑电信号转换为外部设备控制命令的一种系统。论述了共空间模式(CSP)算法在运动想象脑机特征提取的实现过程;提出了在FPGA平台上实现CSP特征提取算法的设计方法,其中包括协方差矩阵计算模块、基于Givens旋转算法的矩阵特征值和特征向量求解模块,并在FPGA开发板上实现了LDA脑电分类;针对本系统对3名实验者进行了测试,分类正确率达到70%以上。实验表明,本设计对今后便携式脑机接口系统的开发具有很好的参考价值。

Brain computer interface （ BCI ） is a system that translating electroencephalogram （ EEG ） into commands toward external peripheral . First , this article discusses the common spatial patterns （ CSP ） feature extraction algorithm implementation process for mo-tor imagery . Secondly , this article proposes the design process for CSP on the FPGA platform , the design includes the covariance matrix calculation module , matrix eigenvalues and eigenvectors solving module based on Givens rotation algorithm . And the LDA classification algorithm for EEG is also realized on FPGA development board . Finally , the system was tested on three experimenters , the correct classification rate was more than 70%. Experiments show that this design has a good reference value for the future de-velopment of portable brain-computer interface systems .