基于改进NBCS的遥感图像特征点匹配算法  

Remote Sensing Image Feature Point Matching Algorithm Based on Improved NBCS

作　　者：吴锐 Wu Rui(Xiamen Urban Construction Municipal Construction Management Co.,Ltd,Xiamen,361026)

机构地区：[1]厦门城建市政建设管理有限公司,厦门361026

出　　处：《福建地质》2022年第1期69-75,共7页Geology of Fujian

摘　　要：特征点匹配是遥感图像配准中的重要环节,但传统的特征点匹配方法存在同名点一对多或者匹配的同名点定位不准确的问题,进而影响匹配精度。针对此问题,在归一化重心坐标系NBCS匹配算法的基础上,在已完成的NBCS特征匹配后进行迭代计算剔除误差过大的同名点对和选择邻域定位更为准确的特征点,经图像特征匹配实验验证,改进算法实现了特征点的准确匹配,提高了匹配精度。Feature point matching is an important link in remote sensing image registration,but the traditional feature point matching method has one-to-many points of the same name or inaccurate location of the same name,which thus affects the matching accuracy.For this problem,on the basis of the normalized coordinates of NBCS matching algorithm,in the completed NBCS feature matching after adding iterative calculation to eliminate the error of the same point and select neighborhood positioning more accurate feature point,the image feature matching experiment,improved algorithm to realize the accuracy of feature point matching,improve the matching accuracy.

关 键 词：特征点匹配 归一化坐标 迭代计算 算法改进 NBCS 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]