论文部分内容阅读
针对立体合成孔径雷达(SAR)影像难以获得可靠匹配的问题,提出一种基于几何约束的立体SAR影像自动匹配方法.该方法首先对多视处理后的SAR影像进行尺度不变特征变换法(SIFT)匹配,并引入几何关系一致性原理剔除误匹配;然后基于构建的几何关系进行格网控制的匹配传播以产生数量更多、分布较均匀的匹配点,并进行归一化互相关(NCC)精化匹配;最后,依据多视视数将匹配点传递到原始影像,在原始影像上基于准确度较高的几何关系进行精化匹配.选择不同视角下的同侧X波段机载立体SAR数据进行试验,与经典SIFT方法对比.结果表明:该方法在匹配效率、匹配精度以及匹配点的空间分布方面具有优越性.
Aiming at the difficulty of obtaining reliable matching of stereo SAR images, an automatic stereo image matching method based on geometric constraints is proposed. Firstly, the scale invariant feature transform (SIFT ), And introduced the principle of geometric relationship consistency to eliminate mismatch. Then, based on the constructed geometric relationship, the mesh-controlled matching propagation was used to generate more and more evenly distributed matching points and normalized cross-correlation (NCC) Finally, the matching point is transmitted to the original image according to the multi-view number, and the matching is performed on the original image based on the accurate geometric relationship.Selecting the same side X-band airborne stereo SAR data from different perspectives The experimental results show that the proposed method is superior to the classical SIFT method in matching efficiency, matching accuracy and spatial distribution of matching points.