Implicit Neural Representations for Deconvolving SAS Images

Synthetic aperture sonar (SAS) image resolution is constrained by waveform bandwidth and array geometry. Specifically, the waveform bandwidth determines a point spread function (PSF) that blurs the locations of point scatterers in the scene. In theory, deconvolving the reconstructed SAS image with the scene PSF restores the original distribution of scatterers and yields sharper reconstructions. However, deconvolution is an ill-posed operation that is highly sensitive to noise. In this work, we leverage implicit neural representations (INRs), shown to be strong priors for the natural image space, to deconvolve SAS images. Importantly, our method does not require training data, as we perform our deconvolution through an analysis-bysynthesis optimization in a self-supervised fashion. We validate our method on simulated SAS data created with a point scattering model and real data captured with an in-air circular SAS. This work is an important first step towards applying neural networks for SAS image deconvolution.

[1]  Brian G Ferguson,et al.  Application of acoustic reflection tomography to sonar imaging. , 2005, The Journal of the Acoustical Society of America.

[2]  John McKay,et al.  Development of an in-air circular synthetic aperture sonar system as an educational tool , 2019, The Journal of the Acoustical Society of America.

[3]  Daniel C. Brown,et al.  Alternative representations and object classification of circular synthetic aperture in-air acoustic data , 2020 .

[4]  L. Lucy An iterative technique for the rectification of observed distributions , 1974 .

[5]  Bernard Mulgrew,et al.  Increasing circular synthetic aperture sonar resolution via adapted wave atoms deconvolution. , 2017, The Journal of the Acoustical Society of America.

[6]  Emmanuelle Gouillart,et al.  scikit-image: image processing in Python , 2014, PeerJ.

[7]  Qinghua Hu,et al.  Neural Blind Deconvolution Using Deep Priors , 2019, 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[8]  Timothy M. Marston,et al.  Coherent and semi-coherent processing of limited-aperture circular synthetic aperture (CSAS) data , 2011, OCEANS'11 MTS/IEEE KONA.

[9]  William H. Richardson,et al.  Bayesian-Based Iterative Method of Image Restoration , 1972 .

[10]  Pratul P. Srinivasan,et al.  NeRF , 2020, ECCV.

[11]  Daniel C. Brown,et al.  GPU Acceleration for Synthetic Aperture Sonar Image Reconstruction , 2020, Global Oceans 2020: Singapore – U.S. Gulf Coast.

[12]  G. L. Bretthorst,et al.  Bayesian Interpolation and Deconvolution , 1992 .

[13]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[14]  Rob Fergus,et al.  Fast Image Deconvolution using Hyper-Laplacian Priors , 2009, NIPS.

[15]  Kyle J Daun,et al.  Deconvolution of axisymmetric flame properties using Tikhonov regularization. , 2006, Applied optics.

[16]  Michael P. Hayes,et al.  Broad-band synthetic aperture sonar , 1992 .

[17]  K. U. Simmer,et al.  A deconvolution algorithm for broadband synthetic aperture data processing , 1994 .

[18]  Hyojin Kim,et al.  Dynamic CT Reconstruction from Limited Views with Implicit Neural Representations and Parametric Motion Fields , 2021, 2021 IEEE/CVF International Conference on Computer Vision (ICCV).

[19]  Jonathan T. Barron,et al.  Fourier Features Let Networks Learn High Frequency Functions in Low Dimensional Domains , 2020, NeurIPS.

[20]  Ce Liu,et al.  Deep Convolutional Neural Network for Image Deconvolution , 2014, NIPS.

[21]  Rob Fergus,et al.  Blind deconvolution using a normalized sparsity measure , 2011, CVPR 2011.

[22]  Andrea Vedaldi,et al.  Deep Image Prior , 2017, International Journal of Computer Vision.

[23]  Jonathan T. Barron,et al.  Deformable Neural Radiance Fields , 2020, ArXiv.

[24]  Kyaw Zaw Lin,et al.  Neural Sparse Voxel Fields , 2020, NeurIPS.

[25]  Brendt Wohlberg,et al.  CoIL: Coordinate-based Internal Learning for Imaging Inverse Problems , 2021, 2102.05181.