Abstract: For the sampled electrocardiogram（ECG） signal，the varied noises often overlay the all frequency field of ECG.It is difficult to use the normal method to de-noising.In this article a non-line threshold filtering function h is introduced.The parameters of the h function in different wavelet scales is determined based on training signals，and then using the adaptive threshold filtering wavelet method to de-noising the ECG signal.Compared with the wavelet threshold de-noising algorithm，this method is validity in ECD de-nosing，and it has more advantages in keeping the ECG shape after de-noising.

摘要： 采集的心电信号，各类噪声往往覆盖了其有用信号的全频段范围，通常的方法难以有效消噪。讨论了将非线性阈值函数h引入小波消噪中，通过训练信号来确定各尺度下的h函数参数，然后采用阈值自适应的小波滤波进行心电信号消噪的方法。通过和Donoho的小波阈值消噪法对实测心电信号消噪比较，说明了该方法在心电消噪方面的有效性，且在消噪后波形不失真方面具有更好的优越性。