Improving the resolution of underwater acoustic image measurement by deconvolution

Improved underwater acoustic imaging with non‐uniform spatial resampling RL deconvolution

“…"#$%&!#"'"("$& )"*"+"$,"#"'"("$& Fig. 1: Acoustic image measurement geometry [28,29] The array is positioned on the xoz plane and parallel to the x axis, the y coordinate of all array elements is 0. The coordinates of the nth array element is ((n − 1)d, 0, h).…”

“…The paper [27] compares the performance of 2D shift-variant RL algorithm in UAI with conventional and MVDR methods. The paper [28] proposes an algorithm for energy correction to improve the performance of the RL algorithm in locating sound sources near the image edge, and compares it with various deconvolution algorithms in UAI. According to [27,28], the RL algorithm has better multi-target resolution and lower sidelobe level than other methods at low SNR.…”

“…The paper [28] proposes an algorithm for energy correction to improve the performance of the RL algorithm in locating sound sources near the image edge and compares it with various deconvolution algorithms in UAI. According to [27, 28], the RL algorithm has better multi‐target resolution and lower sidelobe level than other methods at low SNR. The original 2D RL method has a large computational burden, so a non‐uniform spatial resampling RL fast algorithm is proposed to reduce the amount of calculation.…”

Multiplane deconvolution in underwater acoustics: Simultaneous estimations of source level and position

“…Deconvolution aims to remove the effect of PSF from CBF maps to improve both localization and source level estimation. Approaches such as deconvolution approach for the mapping of acoustic sources (DAMAS) (Brooks and Humphreys, 2006) or non-negative least squares (NNLS) (Ehrenfried and Koop, 2007) have recently been introduced for underwater acoustic imaging in shallow water environments (Sun et al, 2020). These methods consist in solving an inverse problem to find the accurate source distribution (source position and level) considering a given CBF map and a set of predetermined PSFs.…”

“…These methods consist in solving an inverse problem to find the accurate source distribution (source position and level) considering a given CBF map and a set of predetermined PSFs. However, they suffer from heavy computational costs: they require a huge number of PSFs to be computed (as many as the pixel number, several hundreds or thousands) and about one hundred iterations (or more) to converge (Sun et al, 2020). Faster versions of these algorithms require assumptions for shift invariant PSF (Ehrenfried and Koop, 2007) but are not relevant in the deep-sea context: the region of interest can be very large (easily greater than 10 000 m 2 ), with respect to the array dimension (a few meters).…”