摘要
传统基于灰度的影像匹配方法难以适应红外多光谱影像波段间的非线性灰度差异。该文提出一种基于相位一致性的角点筛选与特征匹配方法。首先基于相位最大矩提取影像边缘图,并利用角点的结构属性进行稳定性判断和角点筛选;然后利用相位一致性的幅值和方向信息构建特征描述符,并采用双向最近邻方法实现初始匹配;最后采用RANSAC算法剔除粗差,得到优化后的匹配结果。实验结果表明,该文提出的角点筛选算法能使匹配点中稳定角点的比例由37.9%提升至85.2%,将最大残差由2.765降到1.766,RMSE由0.854降到0.751。与HOPC模板匹配算法相比,该特征匹配算法能在保证匹配精度的同时,将效率提升了3倍。
Traditional gradient-based matching methods cannot adapt to the non-linear radiometric difference between different bands of infrared multi-spectral images.This paper proposes a phase congruency-based corner screening and feature matching method.Firstly,corner structures are extracted from the edge image computed by the maximum moment of phase congruency,and they are used to determine the stability of corners and screen unstable corners.Secondly,feature descriptors of the retained corners are constructed by using the magnitude and orientation of phase congruency.Initial matches are obtained based on the bi-directional nearest neighbor matching method.Finally,initial matches are refined by removing outliers using the RANSAC algorithm.Experimental results show that the proposed method increases the proportion of stable corners from 37.9%to 85.2%,and reduces the maximum residual from 2.765 to 1.766,besides,the RMSE decreases from 0.854 to 0.751.Compared with the HOPC template matching method,the proposed method improves the efficiency by three times while achieving the same matching accuracy.
作者
钟佩珂
江万寿
李金龙
刘晓波
ZHONG Pei-ke;JIANG Wan-shou;LI Jin-long;LIU Xiao-bo(State Key Laboratory of Information Engineering in Surveying,Mapping and Remote Sensing,Wuhan University,Wuhan 430079;Tianjin Jinhang Institute of Technical Physics,Tianjin 300308,China)
出处
《地理与地理信息科学》
CSCD
北大核心
2020年第3期42-48,62,共8页
Geography and Geo-Information Science
基金
国家自然科学基金项目“空间辐射基准传递定标及地基验证技术”(2018YFB0504801)。
关键词
红外多光谱影像
波段匹配
相位一致性
角点筛选
infrared multi-spectral images
band matching
phase congruency
corner screening
