针对基于仿射不变特征的遥感影像匹配技术,提出了一种自动优化方法,以进一步提高匹配准确性。根据典型需求形成了两套优化实施方案,基于所提出的自动优化方法实现了相应具体算法。针对不同类型的多组影像,自动优化的效果与相应方案的预定目标一致,充分证明了本方法的有效性与适用性。
An automated optimization method for affine invariant feature matching on remote sensing imagery is proposed. The correct matching rate is developed as an evaluation criterion of the optimized processing for affine invariant feature matching, which guarantees the quality of the optimized processing and realizes automated processing. Overlap rate is calculated by projecting a local region onto the corresponding region based on homograhpy, upon which a correct matching rate is determined. For different purposes, two classical optimization solutions are introduced. The algorithm for each solution is implemented based on the proposed method. By using one stereo satellite image and two stereo aerial images with different types, the experiment indicates that the results of feature matching can be optimized automated and accurately by our method according to corresponding optimization solution.