Efficient algorithms for acoustic image estimation.

Author(s)

Paulo Otávio Moreira dos Santos

• Document type: Master's Dissertation
• Press: São Paulo.
• Institution: Universidade de São Paulo (USP). Escola Politécnica (EP/BC)
• Defense date: 2021-01-20

Examining board members

Vitor Heloiz Nascimento; Rodrigo Caiado de Lamare; Bruno Sanches Masiero

• Advisor: Vitor Heloiz Nascimento

Abstract

The acoustic imaging problem consists of mapping the directions and intensities of sound sources using a microphone array. Several methods were developed for reducing the computational cost of acoustic image estimation, most of which are based either on the Kronecker Array Transform (KAT) or the non-equispaced fast Fourier transform (NFFT). In this work, we show how to apply the KAT to conventional methods and to more advanced optimization methods. The application of the KAT results in a significant reduction in memory usage and computational cost of a wide variety of methods. We extended the usage of the KAT to new sparse optimization algorithms and developed a new class of homotopy algorithms for acoustic image applications. The KAT was also combined with other efficient methods to solve least squares problems to arrive at very eficient implentations of well stabilished algorithms such as orthogonal matching pursuit (OMP). Finally, we explored methods found in the literatute that were conceptually similar to the KAT to arrive at a new way to estimate acoustic images using Schur-Hadamard products to solve least squares efficiently in conjunction with the KAT. (AU)

• FAPESP's process: 17/22226-1 - Comparison of efficient algorithms for acoustic image estimation
• Grantee: Paulo Otávio Moreira dos Santos
• Support Opportunities: Scholarships in Brazil - Master