Blind separation of acoustic sources from the principle of least spatial complexity

Abstract : A novel algorithm based on the principle of least spatial complexity is proposed to decompose an acoustic field distribution into uncorrelated acoustic sources with maximum spatial compactness. First of all, the acoustic field source distribution is reconstructed by backpropagating the acoustic pressures measured by an array of microphones. In a second step, the reconstructed source field is classically decomposed into uncorrelated sources by means of a principal component analysis. However, it is well-known that having uncorrelated sources is a necessary but by no means a sufficient condition: there exists an infinity of uncorrelated virtual sources that can produce the same acoustic field. In order to find a unique and optimal solution to source separation, it is proposed to invoke a "principle of least spatial complexity" which forces solutions with high spatial compactness. This is achieved thanks to recent results about optimization within the Stiefel manifold. An experiment is conducted to demonstrate the validity of the proposed algorithm. The final result proves that the principle of least spatial complexity and its related algorithm can successfully decompose an acoustic field into its actual constituents.
Complete list of metadatas

Cited literature [9 references]  Display  Hide  Download
Contributor : Bin Dong <>
Submitted on : Thursday, July 24, 2014 - 9:50:29 AM
Last modification on : Tuesday, November 19, 2019 - 10:38:44 AM
Long-term archiving on: Thursday, November 20, 2014 - 4:43:47 PM


Files produced by the author(s)


  • HAL Id : hal-01019695, version 1



Bin Dong, Jérôme Antoni. Blind separation of acoustic sources from the principle of least spatial complexity. International Conference Surveillance 6, Oct 2011, Compiègne, France. ⟨hal-01019695⟩



Record views


Files downloads