基于遥感影像的智能景像匹配适配区选择方法  

Intelligent scene matching suitable area selection method based on remote sensing image

作　　者：范继伟 杨小冈 卢瑞涛 李清格 夏海 FAN Jiwei;YANG Xiaogang;LU Ruitao;LI Qingge;XIA Hai(Missile Engineering Institute,Rocket Force University of Engineering,Xi’an 710025,China)

机构地区：[1]火箭军工程大学导弹工程学院,西安710025

出　　处：《中国惯性技术学报》2023年第1期14-23,共10页Journal of Chinese Inertial Technology

基　　金：国家自然科学基金面上项目(62276274)。

摘　　要：飞行器景像匹配适配区选择是飞行器实现景像匹配视觉导航的前提。近年来,人们基于层次规则提出了许多适配区选择方法,也取得了良好的效果,但这些方法缺少对深度特征的提取能力,通用性较差,且存在一定的误选现象。针对这一问题,提出了一种基于深度特征的智能景像匹配适配区选择方法。所提方法利用深度学习ResNet-50网络结构实现对景像区高维特征的提取,通过深度特征匹配方法计算匹配误差与匹配概率,实现对适配区的选取。实验结果表明,所提方法与传统的适配区选择方法相比,适配成功率平均提高40%以上,鲁棒性更强。该方法避免了繁琐的适配性能指标选择流程,可应用于不同场景下的适配区选择,改善适配区域选择的有效性和泛化性。The selection of aircraft scene matching suitable area is the precondition of scene matching vision navigation.In recent years,many adaptive region selection methods based on hierarchical rules have been proposed,which have also achieved good results.However,these methods lack the ability to extract depth features,have poor versatility,and there is a certain phenomenon of false selection.To solve this problem,an intelligent scene matching suitable area selection method based on depth features is proposed.The proposed method uses the depth learning ResNet-50 network structure to extract high-dimensional features of the scene area.Through the depth feature matching method,the matching error and matching probability are calculated to select the suitable area.The experimental results show that compared with the traditional suitable area selection method,the suitable success rate of the proposed method is increased by more than 40%on average,and the proposed method has stronger robustness.It avoids the tedious process of selecting the adaptation performance indicators,and can be applied to the selection of the suitable area in different scenes to improve the effectiveness and generalization of the selection of the suitable area..

关 键 词：景像匹配 适配区选择 层次规则 深度特征 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]