Imaging Algorithm And Its Implementation On GPU For Ultra-high Resolution Airborne SAR

SAR is a new-style microwave imaging system which could break the resolution limitation of conventional radar and do two-dimensional high resolution imaging on scene. Higher resolution is the unremitting pursuit in SAR development. The SAR image with 0.1m or higher resolution was called ultra-high resolution. Higher resolution can cause longer synthetic aperture and more echo data. In order to weaken the influences which caused by these two changes to the ultra-high resolution SAR imaging, we begin with the following two aspects: on the one hand, from the imaging algorithm perspective, choose an efficient and accurate imaging plan; on the other hand, use the efficient signal processing platform to accelerate the speed of the data processing and realize real-time imaging. Based on the background above, the main work of this paper are as follows:(1) We introduced the conventional PFA imaging algorithms, the two-dimensional interpolation processing for range migration correction is explained and the essence of conversion from polar coordinates to Cartesian format is illustrated from the perspective of signal coupling; introduced the PFA imaging algorithms based on the scale transform and compared the point target simulation results with the same parameters to the conventional PFA imaging algorithms under the same circumstance.(2) We introduced the PGA algorithm and introduced a two-dimensional autofocus algorithm processing scheme with cascade structure based on the PGA algorithms. Overcome blind estimate of two-dimensional phase errors and require the analytical relation between the azimuth phase errors and residual range cell migration under the PFA algorithm signal model. The result of the measured data shows the feasibility of the algorithm.(3) We introduced the GPU's organizational structure and programming technology based on the CUDA environment. An ultra-high resolution SAR imaging algorithm processing scheme was proposed and did researches on its internal parallelism. A GPU parallel acceleration scheme based on this imaging algorithm plan was proposed at the same time. In order to solve the GPU memory restriction, further proposed an asynchronous acceleration optimization scheme. The results of the measured data showed the feasibility and efficiency of the algorithm scheme.