Improving performance of three-dimensional imaging sonars through deconvolution

Abstract

Three-dimensional (3D) imaging sonars are becoming increasingly important for ocean investigation and exploitation. To date, 3D imaging sonars employ a conventional beamforming technique popularly to reconstruct underwater 3D acoustical images. However, angular resolution of the conventional beamforming is limited, which is usually determined by array aperture. Sidelobe levels of the conventional beamforming are also relatively high. These two issues of the conventional beamforming: limited angular resolution and high sidelobe levels limit the performance improvement of 3D imaging sonars. High-resolution imaging algorithms with low sidelobe levels have rarely been investigated in 3D sonars. This paper applies a high-resolution imaging algorithm to improve 3D sonars. We demonstrate that the outputs of the conventional 3D beamforming in a 3D sonar can be formulated as a convolution of the beam pattern of the 2D array employed and the backscattering function of imaging objects, both in the far and near field. Based on the demonstration, the popular deconvolution algorithm in aeroacoustics is applied to deconvolve the conventional beamforming outputs in 3D sonars to improve the spatial resolution and suppress the sidelobes. Both the simulation and experiment verify that using the deconvolution algorithm help improve the angular resolution by more than two times and reduce the average sidelobe level by 20 dB.