Abstract: Wavelet analysis, as one of the best time-frequency analysis methods, has been widely used in signal, figure processing, damage detection and other fields. In this paper, based on binary wavelet transformation and sigmoid function, a new threshold function is build between soft and hard threshold functions. By changing different parameters of this function, it can approximate the soft and hard threshold function in different degrees. According to the filtering results of six different signals by different threshold, the new threshold function can not only improve the signal-noise ratio, but also make the filtered signal more approaching the true signal.

摘要： 小波分析作为一种优秀的时频分析方法，越来越多地应用于信号、图形处理以及故障诊断等领域。在噪声处理领域，基于小波变换的阈值函数较多，但效果不一。基于二进小波变换，在Sigmoid函数基础上，建立介于软硬阈值函数之间新的阈值函数，即通过改变新阈值函数中不同的参数，能够不同程度地逼近软硬阈值函数。从应用不同阈值函数对六种噪声信号滤波效果分析可知，采用改进的阈值函数能够使滤波后的信号在提高信噪比的同时，更接近于真实信号。

关键词: 振动与波, 小波分析, 滤波, 阈值函数, Sigmoid函数