级联双稳随机共振降噪下的有噪ICA研究

Noisy ICA Based on Cascaded Bistable Stochastic Resonance Denoising

研究了强噪声混合条件下的独立分量分析(ICA)问题.提出了一种将级联双稳随机共振(SR)用于有噪ICA盲源分离的方法.该方法利用级联双稳SR对时域波形降噪的优良特性,先对有噪ICA信号进行SR输出,再进行ICA盲源分离.实验结果表明,利用上述方法可以有效提高有噪ICA的分离效果.

The independent component analysis （ICA） in strong noisy mixtures was conducted and a blind source separation method of noisy ICA was presented. This method, based on the cascaded bistable stochastic resonance（SR） , was used to improve the separation performance of noisy ICA by virtue of its denoising ability to time domain signals. The experiment results prove that compared with the situation without denoising, the separation performance after cascaded bistable SR denoising is enhanced greatly.