一种采用小波滤波的独立分量分析算法

在盲信号处理中,作为一种非高斯性度量之一的峭度,对野值可能非常敏感。提出了一种使用小波滤波的独立分量分析(ICA)算法。首先,采用小波对混合信号进行去噪处理。然后,利用主成分分析PCA对混合信号进行白化。最后采用负熵的独立分量分析算法提取混合信号中的独立分量。实验表明,基于小波滤波的ICA算法能有效地提取源混合信号的独立分量,在噪声环境中具有比快速独立分量分析算法更好的盲源分离效果。

Independent component analysis algorithm using wavelet filtering

Generally speaking, signal was interfered within the noise to a certain extent. Besides this, the kurt as one of the measurements of non-Gaussian which may be sensitive to outliers in Blind Signal Processing (BSP) domain.Acording to this two aspects, we proposed an Independent Component Analysis (ICA) algorithm based on Non-entropy in this paper.Firstly, the original signal was denoised by wavelet transform, then the mixed signal was whiten with Principal Component Analysis (PCA) for preprocessing.At last, ...