基于二阶统计的稀疏图像信号盲分离

On the Blind Separation of Sparse Image Signal Based on Second-Order Statistics

盲信号分离是近年来备受关注的信号处理技术,该技术在无需借助任何关于源信号的先验信息的情况下从受扰信号中恢复源信号。稀疏信号的盲分离是盲分离技术的一个重要研究分支,可以利用信号的稀疏性来解决欠定盲分离问题。基于二阶统计的稀疏图像信号盲分离算法,利用二阶矩数字特性判别信号的稀疏性并估计混叠矩阵,从而实现信号的分离,相对于传统的聚类方法,能避免局部收敛问题。

Blind signal separation is a signal processing technique with high concern in recent years, which can recover source signals from disturbed signals without using any prior information of the source signal. The blind separation of sparse image signal, being an important research branch of blind separation technology, can solve the problem of underdetermined blind source separation. The blind separation algorithm of sparse image signal based on second-order statistics, which uses second-order matrix digital features to identify signal sparsity and estimates mixing matrix, can creatively realize signal separation and avoid local convergence problems.