一种改进的星载干涉SAR复图像最大频谱配准算法

最大频谱法常用于星载干涉SAR复图像配准,但该方法计算量较大且易受噪声影响。本文提出一种改进的最大频谱配准算法。该方法利用chirp-z变换替换补零FFT变换,以相对较少的运算量达到较高的频谱峰值计算精度;通过设定合理的判决门限,判定控制点偏移量估计结果的可靠性,以便对位于不同区域的控制点自适应选取子图像截取窗口的长度,达到控制运算量的目的。利用该算法分别对来自ASAR和ERS-1/2的两对复图像进行验证,实验结果表明该算法可以有效实现配准,且比同条件下利用常规最大频谱算法得到的结果更加可靠。

An Improved Maximum Spectrum Peak Co-registration Algorithm for Space-borne InSAR Complex Data

The Maximum Spectrum Peak Algorithm (MSPA) is commonly used in Space-bome SAR complex data co-registration. But the algorithm has large computation load and can be interfered with the noise. An improved co-registration algorithm based on the original MSPA for space-borne SAR complex data is proposed in the paper. It has less computation load and higher accuracy. Chirp-z transform is used to substitute the zero-padded Fast Fourier Transform (FFT) to get a higher accuracy of spectrum peak as well as to control the computation load on a reasonable level. A relative threshold is chosen to judge whether the current estimation of shift between the control points is reliable. Based on the judgment, the control points in different area can choose suitable window length for sub-image cutting adaptive-ly, so that the computation load can be allotted reasonably. The improved algorithm is validated with two pairs of complex images, from ASAR and ERS-1/2 respectively. The experiments results show that the new algorithm has better reliability and lower computation load than the conventional MSPA.