Accelerating range Doppler imaging algorithm for multiple-receiver synthetic aperture sonar on multi-core-based architectures

Abstract

Synthetic aperture sonar (SAS) is an underwater high-resolution imaging method. But with the increase in resolution and mapping width, the amount of raw data used for imaging increases dramatically. To solve the problem of low imaging efficiency of SAS, an acceleration method of SAS imaging in shared memory environment is proposed. By analyzing the calculation characteristics of each step from the original data received to the synthetic aperture imaging result, the range compression, equivalent conversion from multi-receiver signal to single-receiver signal and azimuth compression are designed in parallel with OpenMP instructions, and the multi-core computing resources are fully utilized to accelerate the imaging process. Simulation experiment verifies the correctness of the parallel imaging algorithm. The experimental result of real data shows that the parallel imaging algorithm has high efficiency and can realize super real-time imaging. The efficiency of the proposed method can be changed with the number of computational kernels. The relationship between the acceleration ratio and the computational kernels is approximately linear, which improves the adaptability of the algorithm. Efficient synthetic aperture sonar imaging algorithm provides conditions for post-processing of image, such as image enhancement, image target detection and recognition.