Abstract:

Computation on traditional CPU platforms consumes large amount of time owing to large SAR echo data and imaging algorithm complexity.GPU has the great advantage of many processing cores suitable for the acceleration of independent parallel structure algorithms.With the powerful floating point and parallel processing capabilities of GPU,the ECSA SAR imaging algorithm is verified on the GPU with good results,and the comprehensive (computation time+IO time) speedup has been raised.Since CUDA-GPU utilization will be the future of high resolution spaceborne SAR imaging,the paper gives a preliminary indication of the feasibility of GPU to process spaceborne SAR data,which lays the foundation for further procedure optimization.

Key words: space borne SAR;ECS algorithm;CUDA;GPU