A deep learning method for grid-free localization and quantification of sound sources

作     者：Kujawski, Adam Herold, Gert Sarradj, Ennes 

作者机构：Tech Univ Berlin Einsteinufer 25 D-10587 Berlin Germany 

出 版 物：《JOURNAL OF THE ACOUSTICAL SOCIETY OF AMERICA》 (美国声学会志)

年 卷 期：2019年第146卷第3期

页      面：EL225-EL231页

核心收录：

学科分类：1002[医学-临床医学] 07[理学] 082403[工学-水声工程] 08[工学] 070206[理学-声学] 0824[工学-船舶与海洋工程] 0702[理学-物理学] 

主　　题：Acoustical properties Acoustic beamforming Acoustic signal processing Microphone array Artificial intelligence Optimization algorithms Acoustic noise Artificial neural networks Photoacoustic imaging Functions and mappings 

摘      要：In this contribution it is examined whether the use of deep neural networks can lead to an accurate characterization of single point sources from microphone array data. Based on conventional beamforming maps, the proposed method aims at estimating the source coordinates and the strength. The residual network architecture, a well-established model in the field of image recognition, is successfully applied to this task. The investigation reveals a method that fast and accurately renders the position and strength of an unknown source. Moreover, the accuracy of the position estimation is higher than the grid resolution of the beamforming map.