基于A-DBSCAN的欠定盲源分离算法

为提升欠定盲源分离问题中混合矩阵的估计精度,在噪声环境下基于密度的空间聚类(density-based spatial clustering of applications with noise, DBSCAN)算法的基础上,提出一种自适应确定输入参数的DBSCAN算法(adaptive DBSCAN, A-DBSCAN)用于混合矩阵估计。针对DBSCAN算法邻域半径(Eps)及邻域点数(MinPts)依赖人为设定的问题,首先利用曲线拟合方法得出Eps,然后通过分析聚类输出类别数与噪声点数关系确定MinPts,并将其与混合矩阵估计模型相结合,最后通过最短路径算法实现源信号恢复。实验结果表明,提出的算法在估计混合矩阵和恢复源信号时,相关性能与对比算法相较均有明显提升。

Underdetermined blind source separation algorithm based on A-DBSCAN

In order to improve the estimation accuracy of the mixing matrix in the underdetermined blind source separation, an adaptive density-based spatial clustering of applications with noise (A-DBSCAN) algorithm is proposed, which is based on the DBSCAN algorithm. To solve the problem that the neighborhood radius (Eps) and the number of neighborhood points (MinPts) of the DBSCAN algorithm are determined, First, the curve fitting method is used to obtain Eps, then MinPts is determined by analyzing the relationship between the number of cluster output categories and the number of noise points. The proposed algorithm is combined with the mixing matrix estimation model, and finally the source signal recovery is achieved by the shortest path algorithm. Experimental results show that compared with the comparison algorithm, the proposed algorithm has significantly improved the performance of mixing matrix estimation and source signal recovery.