A method of toeplitz sound source localization based on ransac

Abstract

The invention discloses a Toeplitz sound source localization method based on RANSAC. First, a segment of sound signal X(t) is subjected to frame division processing through a window function w(t) to obtain a frame of sound signal X(t). w (t); Calculate the data covariance matrix R; then perform singular value decomposition on the data covariance matrix R to obtain the signal subspace U s and the diagonal matrix composed of the singular values ​​of the signal subspace using the rotation-invariant subspace idea to construct a matrix Select the first M-1 rows and the last M-1 rows of matrix B, and denote them as B respectively 1 and B 2 , B 1 and B 2 Satisfy the relation: B 1 D=B 2 , where D is the diagonal matrix containing the signal direction information; solve the diagonal matrix D containing the position parameters to obtain the position parameters of the sound signal of the frame; finally, after obtaining the sound source position parameters of each frame, use the RANSAC algorithm to eliminate Unreliable data in these parameters, obtain the optimal sample set and then fit the precise location parameters. The present invention can better fit the effect of positioning parameters and improve the accuracy and robustness of the Toeplitz positioning algorithm.

Description

technical field [0001] The invention belongs to the technical field of sound source localization, and relates to a sound source localization method, in particular to a Toeplitz sound source localization method based on RANSAC. Background technique [0002] Based on the acoustic model of the microphone array, the direction of arrival (DOA) of the sound signal can be obtained by the formula X(t)=AS(t)+N(t). Among them, A is the array manifold of the signal, S(t) is the target signal, and N(t) is the noise signal. The Toeplitz approximation algorithm assumes that the noise signal N(t) is Gaussian noise that is incoherent with the target signal, but in reality, due to the complexity of the propagation environment, there are coherent signal sources in the received signal, including co-channel interference and background objects The multipath propagation signal caused by reflection will cause the accuracy of Toeplitz algorithm to decrease. Contents of the invention [0003] In...

AI Technical Summary

Problems solved by technology

The Toeplitz approximation algorithm assumes that the noise signal N(t) is Gaussian noise that is incoherent with the target signal, but in reality, due to the complexity of the propagation environment, there are coherent signal sources in the received signal, including co-channel interference and background objects The multipath propagation signal caused by reflection will cause the accuracy of Toeplitz algorithm to decrease

Images