Study on Sparse Blinding Separation Based on the Linear Detection and STFT

摘要：

This topic focuses on the underdetermined sparse blinding source separation from the mixtures. A novel algorithm is put forwards based on the linear direction detection and short-time Fourier transformation (STFT). This algorithm includes two stages: (1) the mixing matrix is estimated in the frequency domain by the STFT and a stochastic gradient algorithm; and (2) the linear programming method is applied to separate the source from the mixtures in the frequency and then the frequency domain source is transformed into the time domain source by the inverse STFT. Finally, some experimental results show that the algorithm is effective.

关键词： Sparse blinding separation undetermined the mixing matrix Sparse source stochastic gradient algorithm STFT

作者: YANG Hui-cheng WANG Xiao-xue

作者单位: College of Electrical Engineering, Anhui Polytechnic University, Wuhu, China

会议类型: 国际会议

会议名称: 2010 International Conference on Signal and Information Processing(2010年IEEE信号与信息处理国际会议 ICSIP2010)

会议地点: 长沙

会议语种:英文

页码: 37-41

在线出版日期: 2010-12-14（万方平台首次上网日期，不代表论文的发表时间）