摘要
多通道时频域相干成分提取算法是针对低信噪比的宽频带信号提取问题提出的。它采用多通道同步观测,在各通道的观测数据中信号成分具有较高的相干性,而噪声的相干性较低,因此根据其相干性的高低差别即可将信号与噪声分离,提取有效信号。为实现信号与噪声的分离,首先应用小波包分解将信号在时频域展开,然后通过计算相干系数确定信号的时频分布,最终通过小波包重构将信号从噪声中分离出来。这一算法不需要信号的任何先验知识,收敛快,可以有效地提取宽频带信号,极大地提高信号的信噪比,对非重复性信号具有良好的捕捉能力,应用此算法成功地实现了视觉诱发电位的单次提取。
The multi-channel time-frequency coherence algorithm is presented for extracting wideband signal from strong noises. The algorithm records signal with multi-channel, in which the useful signals in different channels will have higher coherence, while noises have lower coherence. According to this prop-erty the algorithm can separate the signals from noises. First, the recorded noisy signals were expanded in time-frequency domain with wavelet package, then the time-frequency distribution of useful signal was determined according to the coherence, finally the useful signal was reconstructed from the time-frequency distribution. The algorithm can extract wideband signal efficiently with only one iterate and without prior knowledge of the signal. Using this algorithm, extracting video evoked potential with single trial was realized successfully.
出处
《生物物理学报》
CAS
CSCD
北大核心
2003年第3期303-308,共6页
Acta Biophysica Sinica
基金
国家自然科学基金项目(39970204
30170257)
关键词
时频变换
小波包
相干系数
诱发电位
Time-frequency representation
Wavelet package
Coherence
Evoked potential
