Mixing matrix estimation method for unknown number blind separation of sparse sources

Abstract

The invention relates to an alias matrix estimating method of the blind separation of sparse sources with unknown numbers, which belongs to the engineering field and more particularly relates to the technical field of the blind source separation. The method aims at solving the problems that the existing alias matrix estimating method of the blind separation of the sparse sources on the basis of a classic clustering algorithm requires that the number of source signals is known and the estimating precision is poorer. According to the geometric feature that sparse source alias signals present linear clustering and on the basis of the distance relation of a clustering center and each sort of data dense point, the invention provides a novel clustering validity criterion and estimates the number of the source signals according to the criterion. At the same time, each sort of data dense point is found out by making use of Hough transformation so as to substitute the clustering center to estimate the alias matrix, thereby improving the estimating precision of the alias matrix. The method of the invention is suitable for the estimation for the alias matrix of the blind separation of the sparse sources under the condition that the number of the source signals is unknown and is widely applied to the fields of speech recognition, medical signal processing, wireless communication, and so on, and the estimating precision of the alias matrix is improved.

Description

technical field [0001] The invention belongs to the field of engineering, and specifically relates to the technical field of blind source separation. Background technique [0002] Blind Source Separation (BSS) refers to estimating the source signal only from the aliased signal observed by the sensor according to the statistical characteristics of the input source signal without knowing the prior information of the source signal and the transmission channel. Since blind source separation technology does not require prior knowledge of source signals and transmission channels, it has broad practical applications in many fields such as wireless communication, array signal processing, biomedical signal processing, speech signal processing, and image signal processing. prospect. [0003] At present, the most studied blind source separation model is the linear instantaneous mixed model, as shown in the following formula: [0004] x ( t ...

AI Technical Summary

Problems solved by technology

[0011] The present invention aims to solve the problem that the existing aliasing matrix estimation method based on classical clustering algorithm for blind separation of sparse sources requires the number of source signals to be known and the estimation accuracy is poor, and further provides a blind separation of unknown number of sparse sources The aliasing matrix estimation method of

Images