Research Of Scanning Radar Fast Super-resolution Imaging Method

Super-resolution imaging technology based on Scanning radar can obtain angular resolution beyond the limit of real aperture size only by improving signal processing without hardware equipment improvement.It plays an important role in modern microwave remote sensing applications such as surface surveillance,path navigation,weather monitoring,target strike and so on.However,the existing super-resolution methods require high-dimensional matrix operations which can not meet the needs of real-time signal processing in wide-field scenarios.Therefore,it is of great application value to carry out research on the fast super-resolution imaging method of scanning radar.Aiming at the real-time super-resolution imaging requirements of scanning radar in wide-field,the thesis focuses on super-resolution imaging mechanism,fast method bansed on low-rank approximation and signal parallel processing.The main contents are as follows:1.The geometric model and the echo characteristics of super-resolution imaging on scanning radar are researched.According to the beam detection process in scanning mode,the distance history expression is derived and pulse compression and range correction techniques are analyzed.The imaging resolution performance is further discussed which provides the theoretical basis for the later super-resolution processing.2.The fast-super-resolution imaging method based on low-rank approximation is proposed.Through random sampling and low-rank approximation processing,the high-dimensional echo matrix is projected to a low-dimensional linear space,which significantly reduces the computational complexity.At the same time,the adaptive selection of the dimensionality reduction parameters is solved by summarizing the mathematical relationship between the singular values of the matrix and the dimensionality reduction parameters.3.A parallel processing architecture of the super-resolution method on the GPU platform is proposed.The CPU-GPU cooperative working model is constructed according to the parallel computing platform and the CUDA programming principle.The parallel optimization of echo reception and echo processing,preprocessing and super-resolution imaging,and iterative matrix operations are relized by utilizing the hardware architecture of GPU and the independent nature of echo data.The method proposed in this thesis have been verified by simulation experiments and measured data processing.The results show that the proposed fast-super-resolution method enjoys a preferable computational efficiency without performance degradation,which can meet the practical application requirements of real-time imaging.