Research And Implementation Of Parallelization Method For Phase Gradient Autofocus Algorithm For Radar Imaging

Radar imaging is an important milestone in the development of radar technology. It has all-weather, all-weather, long-distance monitoring and positioning characteristics, and can obtain high-resolution radar images similar to optical photography under extremely low visibility meteorological conditions. Optical imaging technology has an irreplaceable advantage. At present, the most widely used imaging radar is Synthetic Aperture Radar, which uses data processing methods to synthesize a smaller real antenna aperture into a larger equivalent antenna aperture, enabling high-resolution radar imaging.Phase Gradient Autofocus(PGA) is the current mainstream self-focusing method, which can effectively improve the focusing accuracy of radar images. It is an important component algorithm in SAR imaging systems. However, the PGA algorithm has the characteristics of cumbersome calculation process, large amount of processing data, and multiple iterations, resulting in a long operation time. Therefore, in order to improve the computational efficiency of the algorithm and reduce the imaging delay of the system, this paper studies the PGA algorithm, especially the two-dimensional block PGA algorithm with better focusing effect, and designs and implements a radar imaging system. PGA parallel processing component.The main research work of this thesis is as follows:1) Combining the UPMM parallel model to analyze the storage access cost of the QorlQ T4240 embedded platform related to the subject, and provide theoretical basis for the parallel data partitioning of PGA algorithm; 2) According to Amdahl's law and Gustafson's law for two-dimensional The parallel performance of the block PGA algorithm is analyzed, and the parallel performance improvement of the algorithm is predicted from the theoretical point of view. 3) Two data partitioning schemes of two-dimensional block PGA algorithm are proposed, and the final scheme is selected by comparison; The requirements of PGA parallel processing components are described, and a lightweight parallel framework based on VxWorks7.0 operating system is designed and implemented. 5) From the perspective of component structure and core function module, PGA parallel processing components are implemented. Detailed design; 6) Implementing components on the QorlQ T4240 highperformance embedded platform with VSIPL computing middleware.For the final implementation of PGA parallel processing components, this paper has carried out detailed testing from the perspective of image data precision and algorithm computing performance. The input and output image comparisons visually show the functional effects of the components, and record the components under different parallelism. The response time indicates that the PGA algorithm does achieve significant performance gains through parallelization.