通信学报 ›› 2021, Vol. 42 ›› Issue (8): 52-60.doi: 10.11959/j.issn.1000-436x.2021140

• 学术论文 • 上一篇    下一篇

基于张量分解的卷积盲源分离方法

马宝泽, 张天骐, 安泽亮, 邓盼   

  1. 重庆邮电大学通信与信息工程学院信号与信息处理重庆市重点实验室，重庆 400065
  • 修回日期:2021-06-30 出版日期:2021-08-25 发布日期:2021-08-01
  • 作者简介:马宝泽（1990- ），男，河北廊坊人，重庆邮电大学博士生，主要研究方向为盲信号分离处理、深度学习
    张天骐（1971- ），男，四川眉山人，博士，重庆邮电大学教授、博士生导师，主要研究方向为盲信号识别、无线通信的智能信号处理、盲信号处理
    安泽亮（1993- ），男，安徽蚌埠人，重庆邮电大学博士生，主要研究方向为调制识别、深度学习、盲信号处理
    邓盼（1990- ），男，四川宜宾人，重庆邮电大学博士生，主要研究方向为信号与信息处理、深度学习
  • 基金资助:
    国家自然科学基金资助项目(61671095);国家自然科学基金资助项目(61371164);信号与信息处理重庆市市级重点实验室建设基金资助项目(CSTC2009CA2003);重庆市教育委员会科研基金资助项目(KJ130524);重庆市教育委员会科研基金资助项目(KJ1600427);重庆市教育委员会科研基金资助项目(KJ1600429)

Convolutive blind source separation method based on tensor decomposition

Baoze MA, Tianqi ZHANG, Zeliang AN, Pan DENG   

  1. Chongqing Key Laboratory of Signal and Information Processing, School of Communication and Information Engineering, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
  • Revised:2021-06-30 Online:2021-08-25 Published:2021-08-01
  • Supported by:
    The National Natural Science Foundation of China(61671095);The National Natural Science Foundation of China(61371164);The Project of Key Laborato-ry of Signal and Information Processing of Chongqing(CSTC2009CA2003);The Research Project of Chongqing Educational Commission(KJ130524);The Research Project of Chongqing Educational Commission(KJ1600427);The Research Project of Chongqing Educational Commission(KJ1600429)

在线阅读

36

PDF下载

534

摘要：

基于张量分解框架提出了一种卷积盲源分离方法，同时解决了混合滤波器矩阵估计和频点排序的问题。首先，根据观测信号的估计自相关矩阵构造出所有频点处的张量模型；然后，利用张量分解技术计算出每个频点上对应的因子矩阵作为该频点的估计混合滤波器矩阵；最后，采用以功率比作为测度的全局优化排序策略消除了全频段的排序模糊性。实验表明，所提方法在不同仿真条件下处理卷积混合的实测语音时表现出了比现有算法更优异的分离性能。

关键词: 卷积盲源分离, 张量分解, 自相关矩阵, 排序模糊

Abstract:

A convolutive blind source separation algorithm was proposed based on tensor decomposition framework, to address the estimation of mixed filter matrix and the permutation alignment of frequency bin simultaneously.Firstly, the tensor models at all frequency bins were constructed according to the estimated autocorrelation matrix of the observed signals.Secondly, the factor matrix corresponding to each frequency bin was calculated by tensor decomposition technique as the estimated mixed filter matrix for that bin.Finally, a global optimal permutation strategy with power ratio as the permutation alignment measure was adopted to eliminate the permutation ambiguity in all the frequency bins.Experimental results demonstrate that the proposed method achieves better separation performance than other existing algorithms when dealing with convolutive mixed speech under different simulation conditions.

Key words: convolutive blind source separation, tensor decomposition, autocorrelation matrix, permutation ambiguity

中图分类号: 

No Suggested Reading articles found!