| 2018 5th International Conference on Advanced Composite Materials and Manufacturing Engineering | |
| The Study of Blind Source Separation Based on Sparsity and Decorrelation | |
| Li, Jiahui^1 | |
| School of Business Management, Shanghai University, Shangda Road 99, China^1 | |
| 关键词: Decomposition algorithm; Decomposition approach; Optimization method; Optimizing functions; Separation performance; Separation process; Sparse algorithms; Time frequency domain; | |
| Others : https://iopscience.iop.org/article/10.1088/1757-899X/394/5/052019/pdf DOI : 10.1088/1757-899X/394/5/052019 |
|
| 来源: IOP | |
PDF
|
|
【 摘 要 】
The main research of blind source separation algorithm is based on sparsity and decorrelation. The traditional sparse algorithm is used to estimate the mixed matrix by sparse signal then the source signal is reconstructed by optimizing function in time-frequency domain. But for some sparse signals, the separation effect is not ideal. The alternate decomposition algorithm can be used as the optimization method for the signal that is less sparse. Alternate decomposition approach is to update the mixed matrix in each iteration and estimate the source signals so that the parameters can be updated in each iteration, the sparse components can be obtained, and then the source signal is reconstructed. If a decorrelation step has been joined in every iteration, it can reduce the correlation between signals effectively, and get sparse and independent of composition in the separation process. It can greatly improve separation performance by this way.
【 预 览 】
| Files | Size | Format | View |
|---|---|---|---|
| The Study of Blind Source Separation Based on Sparsity and Decorrelation | 801KB |
PDF