摘要: 采用2步法研究松弛稀疏性条件下的欠定盲信号分离。在矩阵恢复上,将时域检索平均法从时域扩展到小波域,得到单源区间矩阵恢复算法。在源信号恢复上,分析最短路径法和l1范数算法,提出基于任意观测信号数的统计稀疏分解准则算法。仿真结果表明,相比l1范数解算法,该算法具有较低的计算复杂度,且可提高恢复信号的信噪比。
关键词:
欠定盲信号分离,
单源区间,
松弛稀疏性,
统计稀疏分解准则
Abstract: This paper uses two-step strategy to research underdetermined blind signal separation under relaxed sparsity condition. In matrix recovery, the Searching-and-Averaging Method in Time Domain(SAMTD) algorithm is expanded into the wavelet domain and Matrix Recovery In Single Source Intervals(MRISSI) is obtained. In source recovery, statistically sparse decomposing principle Statistical Sparse Decomposition Principle(SSDP) for arbitrary number of observed signal is proposed after the analysis of the shortest-path method and l1-norm algorithm. Simulation results show that this algorithm has lower computation complexity and it can improve the SNR of the recovered signal compared with l1-norm algorithm.
Key words:
underdetermined Blind Signal Separation(BSS),
Single Source Interval(SSI),
relaxed sparsity,
Statistical Sparse Decomposition Principle(SSDP)
中图分类号:
肖明, 孙功宪, 吕俊. 松弛稀疏性条件下的欠定盲分离[J]. 计算机工程, 2010, 36(12): 274-276.
XIAO Meng, SUN Gong-Xian, LV Dun. Underdetermined Blind Separation Under Relaxed Sparsity Condition[J]. Computer Engineering, 2010, 36(12): 274-276.