概率聚类在非线性信号平滑处理中的应用

本文提出一种基于概率聚类的滤波器。该滤波器通过对活动窗口的输入数据进行聚类,将其聚类中心作为该窗口的输出,从而得到滤波器的输出。从模拟结果我们可以看到该滤波器一方面可以滤去外加噪声,另一方面又保持了输入信号的非线性特征。

Implementing Possibilistic Clustering Approach in Nonlinear Signal Processing Smoothing

In this paper, a filter based on Possibilistic clustering (Possibilistic Clustering Filter, i.. e. PCF) has been proposed. In the moving window, we can get the" good" samples of the data set. Then this filter clusters the input signal's data with PCM (Possibilistic C-mean) algorithm, and uses the center of the cluster as the output of the filter. Through the simulation result, we can see that the filter not only cancels the Gaussian noise, but also presents the nonlinear feature of the input signal.