摘要:
遥感图像配准是图像融合、变化检测、图像镶嵌等应用中不可缺少的步骤。近年来,出现了大量基于尺度不变特征变换( scale invariant feature transform,SIFT)的遥感图像配准方法。在对已有基于SIFT的遥感图像配准文献的调研分析基础上,对各种SIFT遥感图像配准技术进行了系统性评述,将各种改进方法分为SIFT提升算法、针对遥感图像性质的改进、混合方法以及针对算法流程的改进等4类;分析了不同类型改进方法的适用范围和优缺点;指出了SIFT遥感图像配准技术存在的主要问题和难点,对其未来的发展方向进行了展望,为后续研究提供参考。
Abstract:
Remote sensing image registration is an indispensable part in such aspects as remote sensing image fusion, change detection, and image mosaicking. Recently, many SIFT-based remote sensing image registration methods have been presented. In this paper, SIFT - based remote sensing image registration techniques are discussed in a systematic way. Improvements of these techniques are categorized into 4 types, i. e. , application of improved - SIFT, improvement of remote sensing image properties, combination with other methods, and improvement of the algorithm process. Then, the advantages, disadvantages and scope of applications are analyzed for each category. The existent problems and difficulties of SIFT -based remote sensing image registration are summarized. Finally, the prospects of SIFT-based remote sensing registration are predicted so as to provide some valuable references for researchers in this field.
下载: