Improvement of underwater small target sonar images using fast two-dimensional deconvolution

Abstract

The effective detection of underwater small targets largely depends on high-resolution sonar images. Traditional beamformers and matched filters are often used due to their simplicity. However, they have limited angular and range resolution and high sidelobes, which results in insufficient image resolution and image blurring, thus increasing the difficulty of detecting small targets effectively. To improve the sonar system's imaging performance, this paper proposes an underwater small target sonar image enhancement method based on spatiotemporal two-dimensional deconvolution. First, a hyperbolic frequency modulation signal is used as the transmission signal, and the sonar echoes from the far field are received by a uniform linear array. The two-dimensional raw sonar image with azimuth and range information is obtained through traditional sonar imaging methods. Then, spatiotemporal two-dimensional point spread functions are designed, and the modified Richardson-Lucy algorithm is applied along both the angular and range dimensions to deconvolve the raw sonar image. This results in a sonar image with a narrow main lobe and low sidelobes. Finally, simulations and design experiments verify the feasibility of this method. The simulation and experimental results demonstrate that this method can significantly improve target imaging performance, reducing the main lobe width by more than half and suppressing sidelobes to -60 dB. It also exhibits excellent performance in low signal-to-noise ratio (-10 dB) and multi-target scenarios.