High-Resolution Airborne SAR Imaging And Its Implementation On GPU

Airborne Synthetic Aperture Radar(SAR)has many advantages,such as high spatial resolution,large-coverage,the capability of working all time under all-weather conditions and so on,and by now it has been applied widely in many fields.Omega-K algorithm is one of the high-quality SAR processing algorithms.Phase error compensation is necessary to ensure SAR imaging quality.Since phase error is variant in two dimensions,so we need improve the traditional autofocus algorithm to restrain the variance.Due to the rapid development of Graphics Processing Unit(GPU)and the appearance of Compute Unified Device Architecture(CUDA)technology,general purpose GPU(GPGPU)has gained popularity.And GPU provides a new computing platform for SAR imaging.Based on the above considerations,the main assignments of this thesis are as follows:(1)This thesis introduces the imaging model,theory of Omega-K algorithm and motion compensation based on instrument measurement.Also,we analyze the parallelism of the algorithm and design its program based on GPU.The proposed program accelerates point targets imaging by 30 times,which proves that the program is feasible.(2)Since the phase error of stripmap SAR is variant in azimuth,the PGA-MD algorithm is used to estimate phase error.This algorithm takes advantage of the map drift(MD)technique to combine the sub-aperture phase error(SPE)estimated by PGA in a very precise manner.Also,we analyze the parallelism of the PGA-MD and design a solution based on GPU.And the imaging result of measured data indicates the effectiveness of PGA-MD and the feasibility of the parallelism solution.(3)Since the phase error is also variant in range,an improved range segment autofocus technology is proposed.The proposed method segments range cells small blocks,and the neighboring blocks have overlap cells.This method estimates and compensates phase error of every range block by PGA-MD and sub-image PGA.And the sub-image cross-correlating registration technology is proposed to register the neighboring blocks.Then the weighted averaging method is used to mosaic the registered blocks to get the complete imaging result.The imaging result of measured data indicates that range segment autofocus technology can significantly improve the focusing quality.