干涉合成孔径雷达复图像配准解析搜索方法

Analytic Search Registration Approach for InSAR Image

图像配准是干涉SAR数据处理中关键步骤之一,它的处理精度直接影响着高程重建的质量。为了克服经典方法带来误差的影响,本文构建了一种解析代价函数,利用双迭代优化算法与弦割法求取亚像元级偏移量,实现了在连续域内搜索,避免了传统的在离散域搜索带来的误差,提高了配准精度,在插值意义上,配准精度达到最优。基于实测数据的实验结果表明,该算法具有较高精度。

Image registration is one of the crucial steps in InSAR data processing and its processing aceuraney affects the quality of the digital elevation model reconstruction. A subpixel registration approach for accomplishing the registration of InSAR images is presented to overcome the effect of the errors induced by the classical methods. The approach establishes a novel analytic cost function for the extraction of subpixel offsets from corresponding pixels. The biiterative algorithm and the secant method are used to solve the subpixel offsets. Since the proposed approach searches the subpixel offsets in the continuous domain, the registration accuracy is improved, and it achieves a perfect accuracy in a interpolating sense. The experimental result with the real data shows that the algorithm has a high accurancy.