Obstacle segmentation based on the wave equation and deep learning

被引:21
作者
Kahana, Adar [1 ]
Turkel, Eli [1 ]
Dekel, Shai [1 ]
Givoli, Dan [2 ]
机构
[1] Tel Aviv Univ, Dept Appl Math, IL-69978 Tel Aviv, Israel
[2] Technion, Dept Aerosp Engn, IL-32000 Haifa, Israel
关键词
Obstacle identification; Wave equation; Inverse problems; Deep learning; TIME-REVERSAL; INVERSE PROBLEMS; PICKING;
D O I
10.1016/j.jcp.2020.109458
中图分类号
TP39 [计算机的应用];
学科分类号
081203 ; 0835 ;
摘要
We model the inverse physical problem of identifying an underwater obstacle by using the acoustic wave equation. Measurements are collected in a set of sensors placed in the medium. We use this partial information to approximate a segmentation of the obstacle in its correct location. This is an ill-posed problem. We propose a novel deep learning architecture that takes as input the sensor data and computes an approximate segmentation map of the obstacle. (C) 2020 Elsevier Inc. All rights reserved.
引用
收藏
页数:10
相关论文
共 28 条
[1]   Adjoint-weighted equation for inverse problems of incompressible plane-stress elasticity [J].
Albocher, Uri ;
Oberai, Assad A. ;
Barbone, Paul E. ;
Harari, Isaac .
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING, 2009, 198 (30-32) :2412-2420
[2]  
ALLEN RV, 1978, B SEISMOL SOC AM, V68, P1521
[3]   Combined arrival-time imaging and time reversal for scatterer identification [J].
Amitt, Eyal ;
Givoli, Dan ;
Turkel, Eli .
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING, 2017, 313 :279-302
[4]  
[Anonymous], COMBUSTION
[5]  
[Anonymous], 1991, ULTRASON IMAG
[6]   Time-reversed absorbing condition: application to inverse problems [J].
Assous, F. ;
Kray, M. ;
Nataf, F. ;
Turkel, E. .
INVERSE PROBLEMS, 2011, 27 (06)
[7]  
BAER M, 1987, B SEISMOL SOC AM, V77, P1437
[8]   Adjoint-weighted variational formulation for a direct computational solution of an inverse heat conduction problem [J].
Barbone, Paul E. ;
Oberai, Assad A. ;
Harari, Isaac .
INVERSE PROBLEMS, 2007, 23 (06) :2325-2342
[9]  
Bardos C, 2002, ASYMPTOTIC ANAL, V29, P157
[10]  
Baydin AG, 2018, J MACH LEARN RES, V18