Two-dimensional Autofocusing For High-Resolution Airborne SAR And Its Implementation On FPGA

Synthetic Aperture Radar(SAR), as a kind of active detecting system, has been applied and developed extensively for its characteristic of all-time, all-weather, far-distance and large-coverage. Autofocusing technique is of great significance for further phase error estimation and correction to obtain a high-resolution SAR image, since SAR imaging quality is closely related to echo phase. On the other hand, field-programmable gate array(FPGA) become more and more popular in SAR imaging processing systems for its advantages in bulk, treatment efficiency, power dissipation and reliability. This article mainly researched on SAR autofocus algorithms and its FPGA-based implementation.This paper first introduces the background and significance, development history and current situation of SAR autofocus. Then an analysis is made on principle and steps of phase gradient autofocus(PGA) algorithm. Afterward, trying to address the performance degradation of PGA in the presence of significant high-order phase error, which tend to occur when the coherent processing interval(CPI) is long, we discuss a multi-subaperture autofocus algorithm(PGA-MD). The high-order phase error is avoided when full aperture is divided into several segments. The PGA-MD algorithm combines the virtues of PGA and map drift(MD) algorithms, which guarantee the estimating and combining accuracy of the sub-aperture phase error functions. The paper point out the SAR echo error is essentially twodimensional(2-D) later and a 2-D autofocus approach is researched under polar format algorithm(PFA) framework. Based on the analysis of residual error through polar format conversion, the mapping relationship between azimuth one-dimensional(1-D) phase error and 2-D phase error is revealed. And the 2-D estimation problem can be converted to a problem of 1-D estimation/ 2-D correction, accordingly improving the accuracy and efficiency of the algorithm. Subsequently, a block-wise PGA method in image domain is introduced to reduce the impact of space-variant phase error.In order to realize the autofocus processing system on FPGA, we first draw up a 2-D autofocus processing scheme. Then we develop the overall architecture design and separate autofocusing processing system into 3 parts: autofocus algorithms module, data control module and parameters setting module. Requisite explanation and analysis are made on the design procedure of main functional modules such as PGA submodule, Sinc interpolation submodule and cross-correlation submodule, etc, and main-control module. And the mappings of Algorithms onto FPGA are optimized. Next, we expatiate a data storage/transposition module based on DDR3 SDRAM. And we analyze the efficiency of transposition scheme and complete customized interface design. At last introduce the parameters setting module.To confirm validity of the design, we build a verification system mainly based on host computer, Ethernet, FPGA and DDR3 SDRAM. Based on layering model of internet, we illustrate an ethernet-based data communication module, which connects host computer and FPGA board. The paper conducts analysis of system performance from aspects, such as autofocus result, processing accuracy, efficiency and resource utilization. The experimental results show that the FPGA-based twodimensional autofocus processing system can significantly improve qualities of SAR images, and accuracy and real-time capability meet the requirements. At last, after summarizing our work roundly, several topics are brought forward for further study.