Beta 1


Title Sound Source Reconstruction with Covariance Matrix Fitting
Author Lange, Katrine (Scientific Computing, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Supervisor Hansen, Per Christian (Scientific Computing, Institut for Informatik og Matematisk Modellering, Danmarks Tekniske Universitet, DTU, DK-2800 Kgs. Lyngby, Denmark)
Jørgensen, John Bagterp (Scientific Computing, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
Institution Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark
Thesis level Master's thesis
Year 2010
Abstract This thesis is a feasibility study of Covariance Matrix Fitting (CMF) as a Beamforming technique. The main focus will be on improving the quality of the computed solutions by introduction of a regularization term as well as on implementation of an efficient solution algorithm. It is shown how the problem can be formulated as a convex, quadratic programming problem and a reduced formulation is derived to significantly reduce the computational cost of forming the problem. Relevant theory from inverse problems and optimization is presented. The regularization methods Tikhonov regularization and Total Variation regularization are described and the corresponding regularized CMF Beamforming problems are derived. The ill-conditioned nature of the CMF Beamforming problem is discussed and the need of regularization at the lower frequencies documented and sought fullfilled. A primal-dual interior-point algorithm is implemented tailored to solve the Tikhonov regularized CMF Beamforming problem. The algorithm exploits the matrix structure of the problem and a modified search direction is proposed to reduce computation cost. Simulated and real test data are used to illustrate the general quality of the computed solutions as well as how regularization improves the quality of the computed solutions in the severely ill-posed cases.
Imprint Technical University of Denmark (DTU) : Kgs. Lyngby, Denmark
Series IMM-M.Sc.-2010-08
Fulltext
Original PDF ep10_08.pdf (4.00 MB)
Admin Creation date: 2010-04-16    Update date: 2010-10-28    Source: dtu    ID: 259901    Original MXD