Kafaiezadtehrani_Mehrdad.pdf (5.04 MB)
Download file

Under-Determined Blind Source Separation

Download (5.04 MB)
thesis
posted on 24.05.2021, 15:05 by Mehrdad Kafaiezadtehrani
The Under-determined Blind Source Separation problem aims at estimating N source signals, with only a given set of M known mixtures, where M < N. The problem is solved by a two-stage approach. The rst stage is the estimation of the unknown mixing matrix. The contributions made unravel a more precise and accurate tool which directly relates to the initialization of the clustering algorithm. Di erent schemes such as segmentation, correlation and least square curve tting are used to take advantage of the sparsity of the sources. A signi cant addition involves applying linear transforms to produce a higher sparse domain. Further, the second stage is the sparse source recovery using a Matching Pursuit algorithm. The contributions involve a Matching Pursuit algorithm with di

History

Language

eng

Degree

Master of Applied Science

Program

Electrical and Computer Engineering

Granting Institution

Ryerson University

LAC Thesis Type

Thesis