SAR图像最佳欧式空间距离矩阵匹配方法

发布时间：2018-02-25 09:29

  本文关键词： 合成孔径雷达 图像配准 特征向量 尺度不变特征变换　出处：《系统工程与电子技术》2017年05期 　论文类型：期刊论文

【摘要】：在基于尺度不变特征变换算法的合成孔径雷达图像配准算法中,一个特征点通常具有多个主方向,虽然该主方向分配方式可以有效增加正确匹配对数,但是匹配性能会受到特征向量之间的相互影响而下降。文章提出了一种最佳欧式距离匹配方法,该方法通过欧式空间距离矩阵计算待匹配图像两组特征向量集的相似度,获得最佳相似特征点。此外,文章引入代表位置关系的转换距离作为判断特征点空间一致性的依据,有效地消除错误匹配点。与DM等匹配方法相比较,最佳欧式空间距离矩阵匹配方法在匹配精度和匹配效率上验证了其有效性。
[Abstract]:In the synthetic Aperture Radar (SAR) image registration algorithm based on scale-invariant feature transformation, a feature point usually has multiple principal directions, although the allocation of the principal direction can effectively increase the correct matching logarithm. However, the matching performance will be affected by the interaction of the feature vectors. In this paper, an optimal Euclidean distance matching method is proposed, which calculates the similarity between the two groups of feature vectors in the image by Euclidean space distance matrix. The optimal similarity feature points are obtained. In addition, the transformation distance representing the location relationship is introduced as the basis for judging the spatial consistency of feature points, and the error matching points are effectively eliminated, which is compared with DM and other matching methods. The best Euclidean space distance matrix matching method verifies its validity in terms of matching accuracy and matching efficiency.
【作者单位】： 西北工业大学电子信息学院;
【基金】：国家自然科学基金(61401363) 航空科学基金(20155153034)资助课题
【分类号】：TN957.52
，

本文编号：1533874