测绘科学技术学报2025，Vol.41Issue(2)：188-194,7.DOI:10.3969/j.issn.1673-6338.2025.02.011
基于结构特征的多模态遥感图像特征匹配方法
Multimodal Remote Sensing Image Feature Matching Method Based on Structure Feature
摘要
Abstract
The feature matching method using gradient or grayscale information is very sensitive to nonlinear radio-metric differences and geometric differences between multimodal remote sensing images,which makes it difficult to extract similarity features and key points,resulting in unstable or even failed matching.A multimodal remote sens-ing image feature matching method based on structural features is proposed in this paper.Firstly,the original phase consistency structure map is calculated and the spatial structural features of the image are extracted to eliminate the influence of nonlinear radiation differences and geometric differences.Then,a nonlinear scale space is constructed by using a nonlinear diffusion function to detect the feature points of the phase consistency structure map.The 2D-LGF is used to calculate the average amplitude index map and phase consistency direction and generate local fea-ture descriptors to describe the feature points.Finally,the bilateral matching method is used to establish the initial feature correspondence relationship,and progressive filtering is used for convolution operation to restore the true smooth motion field.The consistency of the initially matched motion vectors is used to gradually eliminate false matchings by using iterative method to achieve high-precision matching of multimodal remote sensing images.关键词
多模态遥感图像/结构特征/相位一致性/渐进式滤波/图像匹配Key words
multi-modal remote sensing image/structure feature/phase consistency/progressive filtering/image matching分类
天文与地球科学引用本文复制引用
熊强,柯涛,陶鹏杰,刘小娟,朱康夫..基于结构特征的多模态遥感图像特征匹配方法[J].测绘科学技术学报,2025,41(2):188-194,7.基金项目
湖北省博士后项目(2024HBBHCXB061) （2024HBBHCXB061）
智慧地球重点实验室重点项目(KF2023ZD02-01) （KF2023ZD02-01）
珞珈实验室研究团队基金项目(230700010). （230700010）
