A Research Of Millimeter Wave Radar Super Resolution Method And Hardware Implementation

Based on the scanning radar system,millimeter-wave radar super resolution imaging uses super resolution algorithm to improve the azimuthal resolution of real beam image.It can work in complex environment and bad weather,especially to overcome the blind area of the traditional radar system,complete the imaging of the front area,in the material air drop,aircraft autonomous landing,accurate strike to the ground,automatic obstacle avoidance and other military and civil fields have an urgent demand.Convolution based on super resolution imaging inversion the ill-posedness inherent to obtain,in view of the existing scanning radar echo data,imaging algorithm computation complex problems,a frequency domain was studied from the related iterative super resolution algorithm,design the 3U VPX signal processor architecture,complete algorithm transplantation,set up the system platform,to carry on the field experiment,completed the super resolution real-time processing,specific work content is as follows:(1)Aiming at the inherent ill-health of traditional inverse filtering methods,a frequency-autocorrelation iteration method was used to deconvolution to obtain the target distribution and achieve the azimuthal super resolution.By taking advantage of the Toeplitz structure characteristics of autocorrelation matrix,Levinson Durbin iteration was used instead of matrix inverse to reduce the operation complexity.(2)In view of the large amount of millimeter-wave radar echo data and the high requirement of real-time super resolution imaging,the computation amount of super resolution imaging algorithm was analyzed,and the 3U VPX architecture super resolution signal processor was designed,and the chip selection and board card design were completed.By setting up pipeline structure and ping-pong operation,the super resolution iteration module is divided into 15-level pipelining to maximize the parallelism advantage of FPGA processor and improve the computing speed.(3)The whole super resolution processing system was built,and the main interface modules such as ADC acquisition,DDR3,SRIO and each algorithm module were tested.In addition,field experiments were carried out to verify the super resolution performance and real-time performance of the imaging system.