基于代价函数的盲分离在线白化算法

针对盲信号分离的白化预处理过程，提出了一种基于代价函数的在线白化算法．该算法利用一类代价函数的极值点特性，采用递归计算的方式获得信号的白化矩阵．与传统的白化算法相比，本算法计算量低、利用样本少，因此可适用于信号的在线预白化处理．仿真试验表明，本算法收敛速度快，白化效果理想，为进一步的研究在线盲分离提供了良好的基础．

An On-line Whitening Algorithm for Blind Source Separation Based on Cost Function

The algorithm uses the property of the stationary points of a special cost function and the recursive calculation mode to get the whitening matrix. It has low computing cost and uses only small numbers of samples compared with conventional whitening algorithm, so the algorithm is adaptive to on-line operation.The computer simulations show that the algorithm convergences quickly and can get rich whitening performance, which is a good pretreatment for future on-line blind source separation.