Underdetermined Mixed Matrix Estimation of Single Source Point Detection Based on Noise Threshold Eigenvalue Decomposition

Aiming at the problem that the signal recovery accuracy of the underdetermined blind source separation algorithm is low, the mixed matrix estimation algorithm is improved by using the sparse characteristics of the signal time-frequency domain. By applying the eigenvalue decomposition detection single source point algorithm based on noise threshold to matrix estimation, instead of the single source point detection algorithm of the real time and real part of the traditional time domain, the signal and noise are connected, and the algorithm is improved. Anti-noise performance; then, The k-means algorithm is used to implement the hybrid matrix estimation. Experiments show that the improved algorithm is more accurate than the traditional algorithm under the same conditions, which is more conducive to subsequent signal separation.