Realization of Airborne Forward-looking Radar Super-resolution Algorithm Based on GPU Frame

The deconvolution methods are the effective approaches to realize high cross-range resolution of real-beam scanning radar. However, the computational time of these algorithms cannot meet the real-time requirements of radar imaging, which is a key limitation to apply this technology into engineering application. In order to overcome this problem, we propose a parallel processing plan based on Graphics Processing Unit (GPU) frame to achieve the real-time imaging by the Bayesian deconvolution algorithm. Because of the processing of each range bin is independent, the characteristics of GPU parallel computing can be used to settle the problem of large computation burden. In this paper, the method to realize parallel computing on GPU frame will be explained in detail. At last, the Poisson-based maximum a posteriori (MAP) algorithm [11] is taken as an example to compare the efficiency of the GPU frame with Central Processing Unit (CPU).