基于广义S变换的高斯领域时频滤波方法

针对S变换时频滤波局部时频区域选择问题,提出了一种高斯邻域阈值滤波方法。首先,利用S变换的二维时频特性,构建时频分辨率可调的广义S变换,提高分析信号的时频分辨率。然后根据不同时频分布类型,定义了3种不同的高斯邻域时频滤波模型。针对不同模型高斯邻域,对信号中S矩阵高斯邻域内的时频数据采用局部阈值处理,保留有用信息,通过广义S反变换得到滤波后的信号。仿真与实验结果表明,广义S变换高斯邻域时频阈值滤波方法具有较高的滤波精度,该滤波模型能有效减小滤波均方差。

Novel Gaussian neighborhood time frequency filtering method based on generalized S transform

Aiming at the problem of great error existed in global threshold filtering, a Gaussian neighborhood filtering method based on generalized S transform was proposed. First, a generalized S transformation (ST) was constructed to improve the time frequency resolution. According to the different type, three kinds of different time frequency filtering model were proposed. Confined to the different time frequency models, the time frequency data of S matrix were replaced and dealt with by threshold method, and then filtered signal could be gotten through inverse S transform. Compared with normal ST, generalized S transform has better performance. The simulation results show that the Gaussian neighborhood time frequency filtering method has higher resolution than global threshold filtering, and the proposed model is accurate and effective.