Abstract: The electroencephalogram（EEG） noises possess remarkable outlier in Evoked Potentials（EPS） in  some especial experimentation. The outlier can not be eliminated effectively based on traditional wavelet denoising method. A new method combining median filter and wavelet threshold is proposed. After suppressing the outliers in EPS through median filter, wavelet threshold is used and remained noise is eliminated. The simulation experimental results show that the new method can suppress the outliers in EPs effectively, and it is more robust than traditional wavelet threshold in lower Signal-Noise Ratio（SNR）.

Key words: Evoked Potentials（EPs）, impulsive noise, [α]-stable distribution, median filter, wavelet denoising

摘要： 在一些特殊实验条件下EP信号的背景EEG噪声具有显著脉冲特性。基于传统的小波去噪方法不能有效地去除EP信号中具有尖峰脉冲特性的背景噪声。提出中值滤波-小波阈值去噪算法，进行中值滤波抑制信号中的尖峰脉冲，利用小波阈值消噪去除剩余噪声。仿真结果表明经过中值滤波预处理后的小波去噪方法比传统的小波去噪方法在信噪比较低时更具有良好的消噪性能。

关键词: 诱发电位, 脉冲噪声, [&alpha, ]稳定分布, 中值滤波, 小波去噪