基于支撑集先验的脑电信号正则化子空间重构  

REGULARIZED SUBSPACE RECONSTRUCTION OF EEG SIGNALS BASED ON SUPPORT SET PRIOR

作　　者：杜秀丽[1,2] 张文龙 邱少明 刘庆利[1,2] Du Xiuli;Zhang Wenlong;Qiu Shaoming;Liu Qingli(Key Laboratory of Communication and Network,Dalian University,Dalian 116622,Liaoning,China;College of Information Engineering,Dalian University,Dalian 116622,Liaoning,China)

机构地区：[1]大连大学通信与网络重点实验室,辽宁大连116622 [2]大连大学信息工程学院,辽宁大连116622

出　　处：《计算机应用与软件》2022年第4期105-109,148,共6页Computer Applications and Software

基　　金：辽宁“百千万人才工程”基金项目(2018921080)。

摘　　要：子空间追踪算法(Subspace Pursuit, SP)利用回溯修剪提高了重构准确率,且迭代过程中原子选取更少,复杂度更低,但其性能易受初始支撑集的影响。针对该问题,提出一种基于支撑集先验的多通道脑电信号重构算法。分析了同类别多通道脑电信号支撑集的时空相关性,将同类前一通道的支撑集作为当前通道重构支撑集的先验信息,提升支撑集选取的准确度,进而加快信号重构速度,提高重构的精度。仿真结果表明,在同等采样率下,相较于子空间追踪算法和自适应正则化子空间追踪算法,该算法对多通道脑电信号的重构时间更短,精度更高。Subspace pursuit(SP) improves reconstruction accuracy by using backtracking pruning. Fewer atoms are selected and the complexity is lower in the iterative process. However, its performance is easily affected by the initial support set. To solve this problem, we propose a multi-channel EEG signals reconstruction algorithm based on support set prior. We analyzed the spatio-temporal correlation of the support set for multi-channel EEG signals in the same class. The support set of the previous channel of the same kind was taken as the prior information of the current channel reconstruction support set, which improved the selection accuracy of support set, sped up the signal reconstruction and enhanced the accuracy of reconstruction. The simulation results show that under the same sampling rate, the proposed algorithm has shorter reconstruction time and higher accuracy for multi-channel EEG signals, compared with subspace pursuit and adaptive regularized subspace pursuit.

关 键 词：时空相关性 支撑集 先验信息 重构 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]