Research On Airborne Sliding Spotlight Synthetic Aperture Radar Imaging And Its Implementation On GPU

Synthetic Aperture Radar(SAR)is a microwave imaging system,which has high imaging resolution and long range of action,works all day and all weather,and plays a huge role in military and civilian fields by virtue of its many advantages.In the context of continuous improvement in the application of synthetic aperture radar technology,the synthetic aperture radar has developed many working modes such as Strip-map model,Spotlight mode,and Sliding Spotlight mode,Scan model,Terrain Observation by Progressive Scans SAR.The sliding spotlight mode not only solves the problem of limited positional resolution under the strip mode,but also solves the problem of limited imaging scene size in the spotlight mode.Between the imaging resolution and the imaging scene,the sliding spotlight mode balances the advantages and disadvantages of the two.Therefore,sliding bunching has higher research value and practical value.However,the SAR synthetic aperture in the airborne sliding spotlight mode requires a longer time,a larger amount of data,and a larger amount of calculation.Data processing on traditional CPU-based workstations or CPU-based servers will consume a lot of time and cannot adapt to real-time Sexually demanding occasions.So its required to continue to explore new signal processing technologies to meet real-time requirements.Since NVIDIA launched the GPU(Graphic Processing Unit)in 1999,after 22 years of continuous rapid development,the use of GPU has broken through the traditional image processing and display fields.In 2007,NVIDIA launched the CUDA-based(Compute Unified Device Archetecture)'s new architecture GPU,GPU's powerful high-speed parallel computing capabilities allow scientific and technological workers around the world to increase the scientific computing research on GPU algorithm speed.Researchers of SAR have also noticed the advantages of this technology,and try to apply the powerful computing power of GPU to SAR imaging algorithms to shorten the imaging time of SAR,in anticipation of solving the problem of SAR's high real-time requirements.Based on the above background,the main tasks completed in this paper are as follows:(1)The differences and connections between the spotlight mode,the strip mode and the sliding spotlight mode are analyzed and compared according to the imaging geometric model,distance migration,azimuth resolution and azimuth Doppler frequency,etc.Then their respective advantages and disadvantages are summarized.At the same time,the airborne sliding spotlight SAR imaging model is analyzed and established,and based on this model,the SAR signal echo equation in sliding spotlight mode is derived.(2)Analyzed the traditional imaging algorithm,and aiming at the spectrum aliasing caused by the low azimuth sampling rate of the sliding spotlight SAR signal,the sliding spotlight mode SAR imaging algorithm is researched in this paper,which systematically studies the up-sampling imaging algorithm and the azimuth based-band scaling algorithm(Based-band Azimuth Scaling Algorithm,BAS),and detailed theoretical derivations and signal simulation verifications of the latter algorithm are carried out.Finally,this imaging algorithm is tested with the measured data of the ??? Institute.(3)The organization structure and programming technology of GPU based on CUDA environment are introduced.The inherent parallelism of the airborne sliding spotlight SAR imaging processing scheme is studied,and a parallel implementation of the imaging processing scheme is given.In view of the larger SAR processing data volume of the measured data and the limited GPU memory,a GPU accelerated processing optimization scheme based on asynchronous processing is proposed on the basis of the above scheme.In the end,the effectiveness of the BAS imaging algorithm is verified by the actual measurement of the airborne sliding spotlight data processing results of the ??? Institute,and it fully reflects the improvement of imaging efficiency brought by GPU parallel acceleration,which provides research ideas for solving the real-time problem of radar measured data processing.