基于稀疏表示和外输入自回归模型的单次诱发电位提取方法

发布时间：2018-05-27 14:28

本文选题：诱发电位 + 稀疏表示　；参考：《信号处理》2017年S1期

【摘要】：由于诱发电位信号的信噪比很小,因此诱发电位的单次提取一直是脑电信号处理领域的难题之一。稀疏表示在信号去噪方面是一个强大的工具,并且诱发电位已被证明在合适的稀疏字典上具有很强的稀疏性。本文提出了一种基于稀疏表示和外输入自回归模型的单次诱发电位提取算法。其中,诱发脑电信号提取的过程分为三个阶段。首先,该算法应用参考信号估计自发脑电的自回归模型参数;其次,在自回归移动平均模型中,应用稀疏表示替代移动平均模型对诱发电位进行建模;最后,利用稀疏系数和自回归模型参数重构诱发电位。通过仿真脑电信号和真实诱发电位信号进行实验验证,结果表明该算法具有较好的效果,与MOSCA和ARX方法相比,本文算法能够在低信噪比情况下进行潜伏期和幅度的估计。
[Abstract]:Because the signal-to-noise ratio of evoked potential signal is very small, the single extraction of evoked potential is always one of the difficult problems in the field of EEG signal processing. Sparse representation is a powerful tool for signal denoising and evoked potentials have been proved to be highly sparse in a suitable sparse dictionary. In this paper, a single evoked potential extraction algorithm based on sparse representation and external input autoregressive model is proposed. Among them, the process of evoked EEG signal extraction is divided into three stages. First, the algorithm uses reference signal to estimate the parameters of autoregressive model of spontaneous EEG; secondly, in the autoregressive moving average model, sparse representation is used to model evoked potential instead of moving average model. The sparsity coefficient and the parameters of the autoregressive model were used to reconstruct the evoked potential. The experimental results show that the proposed algorithm is effective, and compared with the MOSCA and ARX methods, the proposed algorithm can estimate the latency and amplitude at low signal-to-noise ratio (SNR).
【作者单位】：江苏师范大学电气工程及自动化学院;
【基金】：国家自然科学基金(61401181) 徐州市科技项目(KC16SY160)
【分类号】：R318;TN911.7

【相似文献】