基于分数低阶ICA的混合噪声中脉冲噪声的消除方法

Elimination of pulse-noise from mixed-noise based on fractional lower order ICA

为了解决当信号中同时含有高斯噪声与α稳定分布脉冲噪声时,传统的基于二阶统计量的去噪方法会出现显著的性能退化的问题。通过基于负熵的分数低阶独立分量分析(ICA)算法先去除混合噪声中的脉冲噪声,然后利用SVD算法再去除混合信号中高斯噪声的方法,取得了较好的效果,且有效保护了混合信号中的纯净信号。

In order to solve the problem that traditional denoising methods based on the second order statistics were usually degraded in their performance when the received signal was contaminated by both Gaussian and Alpha stable noises.A fractional lower order statistics based ICA method was introduced.It eliminated the pulse noise from the mixed noise first,and then removed the Gaussian noise with the SVD method.The simulations show that the new method is effective on the noise elimination,and the pure signal in the received signal is not deteriorated.