中文摘要：

目的鉴于匹配追踪算法具有良好的参数化描述特性，应用匹配追踪算法研究癫痫脑电的时频分布特征。方法通过仿真算例，将匹配追踪算法与短时傅里叶变换、Wigner—Viiie分布结果进行比较，验证该方法的频率分辨率高及参数化表征的优越性；应用上述3种方法对癫痫脑电和正常脑电进行时频分析，研究癫痫异常放电在时频平面的表现。结果仿真结果表明基于匹配追踪算法能得到较好的时频分布；对癫痫脑电和正常脑电进行时频分析，癫痫脑电和正常脑电在时频平面上存在明显的差异。结论基于匹配追踪的时频分析方法，能够更好地揭示脑电类非平稳信号的特征。

英文摘要：

Objective Matching pursuit algorithm（MAP）, for its good parametric characterization, was applied in epileptic electroeneephalography（EEG） to study time-frequency distribution. Methods Simulation experiment of time-frequency analysis was carried out to verify the matching pursuit algorithm＇s superiority on frequency resolution and parametric characterization. Fourier transform, Wigner-Ville distribution and matching pursuit algorithm were applied to the time-frequency analysis on normal EEG and epileptic EEG to study epileptic discharge in the time-frequency plane and the results were compared. Results Simulation results showed that the matching pursuit algorithm obtained a better time-frequency distribution. Distributions of epileptic EEG and normal EEG had significant difference in time-frequency plane. Conclusion Time-frequency analysis based on matching pursuit can better reveal the EEG characteristics.