应用于SAR图像配准的自适应SIFT特征均匀分布算法

发布时间：2018-04-03 08:28

本文选题：图像匹配　切入点：SIFT特征　出处：《武汉大学学报(信息科学版)》2015年02期

【摘要】：SAR图像配准要求提取出稳定且分布均匀的同名点,但是传统的SIFT算法没有考虑特征点的空间分布情况,提取的特征呈块状分布。针对SAR图像数据的特点,提出了一种自适应控制SIFT特征均匀分布算法,利用局部纹理特征,并结合最优化筛选策略,在保证特征点稳定性和准确性的同时,自适应控制特征在不同空间的分布情况,实现SIFT特征点在图像空间和尺度空间的合理分布。最后从提取效率、匹配正确率和分布质量等方面进行对比试验,验证了算法应用于SAR图像配准的能力。
[Abstract]:SAR image registration requires the extraction of stable and uniformly distributed points of the same name, but the traditional SIFT algorithm does not consider the spatial distribution of feature points, and the extracted features are block distribution.According to the characteristics of SAR image data, an adaptive control SIFT feature uniform distribution algorithm is proposed, which utilizes local texture features and combines with the optimal screening strategy to ensure the stability and accuracy of feature points at the same time.The distribution of adaptive control features in different space can realize the reasonable distribution of SIFT feature points in image space and scale space.Finally, a comparative experiment is carried out in the aspects of extraction efficiency, matching accuracy and distribution quality, which verifies the ability of the algorithm to apply to SAR image registration.
【作者单位】：中国科学院空间信息处理与应用系统技术重点实验室;中国科学院电子学研究所;中国科学院大学;
【基金】：国家863计划资助项目(2012AA121304)~~
【分类号】：TN957.52

【相似文献】