一种基于组合特征的异源遥感影像配准方法

A Multi-Feature-Based Registration Method Adapted to Multi-Source Remote Sensing Images

为实现大差异光学影像的高精度配准,充分利用各类特征检测算法的优势,提出一种基于多类型特征组合的异源遥感影像配准方法。在对获取的影像特征进行描述的基础上,通过影像分块均匀性分析和RANSAC整体一致性分析筛选高精度可靠的组合特征,构建和解算综合参数模型以实现影像的预配准和重采样。再利用最小二乘方法优化特征点匹配结果,使用小面元的方法进行影像的精配准和重采样。实验结果表明在异源光学遥感影像的亮度、分辨率差异较大时,可得到高精度的配准结果。

To complement the advantages of different feature extraction algorithms during the high accurate registration process of multi-source images,a registration method using multi-feature fusion is proposed in this paper.In this method,random sample consensus and uniformity analysis are used to obtain high quality control points.And the integrated parametric model is built for coarse registration.Least squares optimization and rectification with tiny facet primitive are used for precise registration.The experiment confirmed that the proposed method can get a stable result with high precision under the condition of different levels of image brightness or resolutions.