基于小波平滑的超高斯与亚高斯信号盲源分离算法

Blind Source Separation Algorithm Based on Wavelet Smoothing for Super-Gaussian and Sub-Gaussian Signals

为了分离超高斯与亚高斯信号,利用小波变换的高低频系数作为平滑因子,建立以分母作为预测误差的信噪比目标函数,优化目标函数以求解分离矩阵.仿真表明,该算法能够有效地分离出源信号.

In order to separate super-Gaussian and sub-Gaussian signals, this paper uses highand low-frequency coefficients of wavelet transform as smooth factors, then builds a signal-to-noise ratio objective function, which uses the denominator as prediction error and can be optimized to resolve separable matrix. Simulation shows that this algorithm can separate source signals effectively.