聚束SAR快速卷积反投影成像算法的研究

Research on Fast Convolution Backprojection Algorithm for Spotlight Synthetic Aperture Radar

对应用于聚束合成孔径雷达成像的卷积反投影(CBP)算法进行了研究,提出了一种新的快速反投影算法,在分割图像及角度降采样基础上,采用递归调用方法,降低了CBP算法的计算量,相对于直接反投影的计算量O(N3),快速反投影算法的计算复杂度可降低为O(N21bN)。同时讨论了递归参数和角过采样参数的选择对算法的影响,仿真了四种插值方法对运算性能的影响,从而分析了计算复杂度和精确度的关系,并且通过与直接反投影算法仿真结果的比较,验证了快速卷积反投影算法在较小均方根误差水平下可以显著减小计算量。

In this paper, one of spotlight SAR imaging algorithms, the convolution backprojection （ CBP） algorithm, is studied in detail. A new fast backprojection reconstruction algorithm for spotlight-mode SAR is proposed to reduce computation load. Based on the decomposition of the image and angular downsampling, the new algorithm adopts a recursive method and thus reduces the computational cost. The choice of the parameters that affect the behavior of the algorithm, such as recursion and angular oversampling is discussed, and the effects of four different interpolation methods are simulated. The analysis of the relationship between cost and distortion is also presented. Comparison with direct CBP suggests that FCBP offers acceptable root mean square reconstruction errors with considerable speedup