一种宽幅星载TOPSAR数据高效成像算法

An efficient imaging algorithm for spaceborne TOPSAR data

得益于天线波束方位向扫描技术,星载合成孔径雷达(SAR)能够在渐进扫描模式(TOPS)下实现宽测绘带观测.在TOPS模式下,SAR波束沿着方位向进行逆时针反复扫描,并在距离向上进行多次切换,从而获取多个图像切片(burst).为了获得均匀采样的完整成像结果,通常需要对各burst进行重采样,但这会增加额外的计算量,降低处理效率.本文提出了一种宽幅星载TOPSAR数据高效成像算法.首先,在去斜率操作中引入一种新的插值方法,可以灵活地调整最终图像的方位采样间隔,方便进行图像拼接.该方法避免了在聚焦处理前的补零操作,从而降低计算量.在此基础上,本文结合TOPS模式多burst的特点,设计了一种GPU并行处理架构,成倍地提高数据处理效率.最后,根据仿真数据进行了实验,验证了该成像算法的有效性和高效性.

By steering antenna beam in along-track direction,space-borne synthetic aperture radar(SAR) can achieve wide swath coverage in Terrain Observation by Progressive Scans(TOPS) mode.In TOPS mode,the beam switches in range and steers along azimuth several times to perform wider swath both in range and azimuth.In order to obtain a whole imaging result with uniform azimuth intervals,the re-sampling operation is usually required,which means large computation load and low processing efficiency.In this paper,a modified three-step algorithm based on Graphics Processing Units(GPU) is proposed.Firstly,a new interpolation method is introduced into the de-ramp operation(the third step of the three-step focusing algorithm),which avoids the up-sampling operation and can conveniently adjust the azimuth interval of final images.Then,a parallel framework of GPU is designed to accelerate the imaging processing exponentially.Finally,experiments based on simulation data are conducted to verify the validness and efficiency of the imaging algorithm.