非刚性点集匹配的机载下视图像目标定位方法  

Airborne down-view image target location method with non-rigid point set matching

作　　者：袁东华 赵化启[1] 王宇春[1] 程岩 郭浩 刘琳 李国晶[2] 刘晓敏[1] YUAN Donghua;ZHAO Huaqi;WANG Yuchun;CHENG Yan;GUO Hao;LIU Lin;LI Guojing;LIU Xiaomin(College of Information and Electronic Technology,Jiamusi University,Jiamusi Heilongjiang 154000,China;College of Materials Science and Engineering,Jiamusi University,Jiamusi Heilongjiang 154007,China)

机构地区：[1]佳木斯大学信息电子技术学院,黑龙江佳木斯154000 [2]佳木斯大学材料科学与工程学院,黑龙江佳木斯154007

出　　处：《智能计算机与应用》2022年第12期44-50,共7页Intelligent Computer and Applications

基　　金：黑龙江省省属高等学校基本科研业务费项目(2019-KYYWF-1384);黑龙江省卫生健康委立项科研课题(2019-287);佳木斯大学优秀学科团队项目(JDXKTD-2019008);佳木斯大学教育教学改革研究项目(2021JY2-02);佳木斯大学教育教学改革研究项目(2021JY1-49)。

摘　　要：机载下视目标定位在航空摄影和导航等领域有着广泛的应用,其定义为给定目标图像确定目标在参考图像中的位置。而机载下视目标定位中目标图像与参考图像之间存在的大尺度形变问题是有待攻克的难点问题,针对此问题本文提出了一种非刚性点集匹配的目标定位方法。首先,使用SIFT(Scale-Invariant Feature Transform)算法提取2幅图像的关键点,随机K-D树最近邻算法用于获得一致性关键点集;然后,使用K-means分类方法将一致性关键点集划分成多个子集,并在多个子集上使用随机采样微分同胚点集匹配方法确定空间变换模型,解决了大尺度形变问题;最后,通过SSD(Sum of Squared Difference)来确定最优目标位置。实验结果表明所提方法能够有效完成目标定位任务。Airborne down-view target localization has a wide range of applications in the fields of aerial photography and navigation, which is defined as determining the position of a target in a reference image of a given target image. The problem of large scale deformation between the target image and the reference image in airborne down-view target localization is a difficult problem to be overcome, and a non-rigid point set matching target localization method is proposed in this paper to address this problem. Firstly, SIFT(Scale-Invariant Feature Transform) algorithm is used to extract the key points of the two images, and randomized K-D tree nearest neighbor algorithm is used to obtain the consistent key points set. Then, the consistent key point set is divided into multiple subsets using the K-means classification method, and the spatial transformation model is determined using the random sampling differential homozygous point set matching method on multiple subsets to solve the large scale deformation problem. Finally, the optimal target location is determined by the SSD(Sum of Squared Difference). The experimental results show that the proposed method can effectively accomplish the target localization task.

关 键 词：目标定位 关键点检测 点集匹配 微分同胚 随机采样 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]