论文部分内容阅读
配准是合成孔径雷达干涉测量(In SAR)得到干涉图前的关键步骤。本文提出了基于SIFT和RANSAC的In SAR影像配准。尺度不变特征变换(SIFT)和随机采样一致性(RANSAC)相结合,得到特征稳定、匹配点对可靠的In SAR影像配准。利用PALSAR数据及Terra SAR-X数据,进行不同分辨率、不同波段的SAR影像配准试验,结合生成的干涉图条纹的清晰程度,评价其精度。并与目前主流的交叉互相关方法得到的相干系数进行了对比,证明了该方法在In SAR影像配准应用中是一种简单、有效的方法。
Registration is a key step before SAR interferometry (In SAR) is achieved. In this paper, InSAR image registration based on SIFT and RANSAC is proposed. SIFT and RANSAC are combined to obtain a stable feature point and a reliable matching point to In SAR image registration. The PALSAR data and Terra SAR-X data were used to carry out SAR image registration experiments with different resolutions and different bands. The accuracy of the generated interference pattern fringes was evaluated based on the clarity of the generated interference pattern fringes. Compared with the correlation coefficients obtained by the current mainstream cross-correlation method, it is proved that this method is a simple and effective method for In SAR image registration.