盲信源分离中信源动态变化的识别  

Blind source separation for identifying the changing source number

作　　者：刘小东[1] 楼顺天[1] 

机构地区：[1]西安电子科技大学电子工程学院,陕西西安710071

出　　处：《西安电子科技大学学报》2003年第5期598-602,共5页Journal of Xidian University

基　　金：国家自然科学基金资助项目(20072043)

摘　　要：基于独立分量分析(ICA)的盲信源分离技术以及小波包分解技术,提出了根据能量特征向量确定信源数的方法,定义了各输出分量的能量特征,并根据输出分量能量特征向量之间的距离,检测出冗余的输出信号,进而确定出信源数.计算机仿真结果验证了其有效性.Based on the energy eigenvector, a new approach for source number determination is presented by using the Independent Component Analysis (ICA) and the wavelet packet decomposition arithmetic. The energy eigenvector of output components is defined. According to the distance among the output component energy eigenvector the redundance output components can be detected and the number of source can be determined. The effectiveness of the arithmetic presented in this paper is verified through the computer simulation.

关 键 词：独立分量分析 盲信源分离 小波包分解 自然梯度算法 

分 类 号：TN911.7[电子电信—通信与信息系统]