论文部分内容阅读
提出了一种基于多通道匹配追踪算法对一维多通道脑磁图非平稳信号进行单次分解的方法.单通道匹配追踪算法通过在过完备的库中自适应地搜索匹配具有良好时频特性的Gabor原子,最终将信号表示为Gabor原子的线性组合.多通道匹配追踪算法是单通道匹配追踪算法的延伸,即所有通道的信号是由发生时间、振荡频率、相位、持续时间相同,但振幅不同的Gabor原子线性叠加组成.结果显示,多通道匹配追踪算法能较好地重建源信号,并且自动检测与分离伪迹噪声.此外,3D源重建结果和心理学先验知识一致,这进一步表明多通道匹配追踪算法应用于脑磁信号单次提取是有效的.
A method based on multi-channel matching pursuit algorithm for one-dimensional decomposition of non-stationary one-dimensional multi-channel magnetoencephalography signals is proposed.A single-channel matching pursuit algorithm is adaptively searched and matched in an overcomplete library with good time-frequency characteristics Gabor atoms, and finally represent the signal as a linear combination of Gabor atoms.Multi-channel matching pursuit algorithm is an extension of single-channel matching pursuit algorithm, that is, the signals of all channels are the same in terms of time of occurrence, oscillation frequency, phase and duration, but the amplitude The results show that the multi-channel matching pursuit algorithm can reconstruct the source signals well and automatically detect and separate the artifact noise.In addition, the 3D source reconstruction results are consistent with the prior knowledge of psychology, which further shows that Multi-channel matching pursuit algorithm applied to brain magnetic signal single extraction is effective.