论文部分内容阅读
针对现有的欠定盲分离混合矩阵估计方法中存在的估计精度低以及时间复杂度高等缺点,提出一种基于相似度检测的欠定混合矩阵估计方法,该方法能够在没有任何先验信息的条件下自适应地估计出源信号数目以及混合矩阵,而且不需要进行迭代,时间复杂度低.仿真结果表明,与现有的一些混合矩阵估计方法(如改进K均值聚方法和拉普拉斯势函数法)相比,所提出的方法在源信号数目估计准确率、混合矩阵估计精度以及时间复杂度等方面都具有明显优势.
Aiming at the disadvantages of low estimation accuracy and high time complexity in the existing undefined blind separation hybrid matrix estimation methods, an under-determined mixed matrix estimation method based on similarity detection is proposed. This method can be used in the estimation of mixed matrix without any prior information The number of source signals and the mixing matrix are adaptively estimated without iteration and the time complexity is low.The simulation results show that compared with the existing hybrid matrix estimation methods such as improved K-means clustering method and Laplacian Compared with the proposed method, the proposed method has obvious advantages in the accuracy of the estimation of the number of source signals, the accuracy of the hybrid matrix estimation and the time complexity.