Study On Arbitrary Track SAR Imaging Processor Based On GPU

Synthetic Aperture Radar(SAR)has the advantages of working all day with long working distance,little influence of environmental impact,and high imaging precision.It is widely used in civil facilities and military equipment.With the increasing accuracy of imaging and the complexity of radar trajectory,it is urgent to propose a new and efficient radar imaging algorithm.In this paper,the SAR imaging method and its parallel design based on the Graphics Processing Unit(GPU)are studied in this paper,including the Back Projection(BP)imaging algorithm and its improved algorithm.The fast algorithm of BP based on arbitrary trajectory is also explored.Then,the FFBP algorithm of any trajectory is designed based on GPU,which effectively improves the efficiency of the algorithm and solves the problem that the FFBP algorithm consumes long time in practical engineering applications.The main contents of this paper are as follows:1.A brief introduction to the parallel programming of the two modes,multi-core CPU-based multi-threaded parallel and GPU-based multi-threaded parallel mode.The GPU is described in detail,and the GPU development platform(CUDA)is described in detail,which will lay a foundation for the parallel design of the FFBP algorithm.2.The SAR echo model and SAR imaging are discussed in detail.Including the specific form of SAR echo,the principle of pulse compression,the principle of two-dimensional high-resolution,and the most classic and basic imaging algorithm Range-Doppler algorithm etc.,which provides theoretical basis for later imaging process.3.The SAR imaging processor based on arbitrary trajectory is studied,and the basic principle of the imaging algorithm-BP imaging algorithm and its improved optimization algorithm is introduced.Then the FFBP algorithm of arbitrary trajectory is designed and optimized,and finally completed the FFBP algorithm GPU implementation.4.Finally,the FFBP algorithm based on arbitrary trajectory is experimented with SAR echo data of linear trajectory and curve trajectory respectively.The effectiveness of the algorithm and its parallel design is verified.Compared with the Matlab based on CPU,the GPU based on CUDA is faster when processing the same data.