摘要
论述了负熵最大化的基本原理和判断条件,在此基础上建立了一种基于负熵、通过数学迭代的方法得到的快速定点抽取算法。该算法具有迭代次数少的显著优点。通过仿真验证了算法的有效性,并将该算法应用到实际语音信号的盲分离实验中。实验结果表明,所建立的算法对盲源分离具有优良的性能,通过与基于峭度的算法对比,发现该算法具有很好的鲁棒性。
This paper discusses the fundamental,discrimination condition and practical algorithm of negentropy maximization,founds the fast fixed-point extract algorithm based on the negentropy and adopts the math iterate algorithm.The algorithm has a prominent excellence of very little iterative times.The effectiveness of the algorithm is confirmed and the results of the experiments are given through the simulation and the algorithm is applied to really speech signal′s experiment of blind source separation.The results of experiment show that this method has good performance for blind source separation.This method have good robustness through comparing with the algorithm based on the kurtosis.
出处
《常州工学院学报》
2010年第1期24-27,共4页
Journal of Changzhou Institute of Technology
关键词
独立分量分析
盲源信号分离
负熵
快速定点抽取
negentropy maximization
blind source separation
speech signal
fast fixed-point extract algorithm