一种基于多目视觉的无人机实时定位方法  

A Real-time Unmanned Aerial Vehicle Localization Method Based on Multi-view Stereo Vision

作　　者：王辰 郑顺义[1] 朱锋博 桂力[1] WANG Chen;ZHENG Shunyi;ZHU Fengbo;GUI Li(School of Remote Sensing and Information Engineering, Wuhan University, Wuhan 430079, Chin)

机构地区：[1]武汉大学遥感信息工程学院,湖北武汉430079

出　　处：《测绘通报》2018年第4期32-35,82,共5页Bulletin of Surveying and Mapping

基　　金：中央高校基本科研业务费专项资金项目(2042016kf0012);湖北省科技支撑计划项目(2015BCE080)

摘　　要：提出了一种基于多目视觉的实时无人机自动定位方法。该方法将多相机和多台计算机联合组建局域网,通过相机标定、图像同步采集、背景差分与前方交会等实现无人机自动检测与定位。为了验证该方法的有效性,开发了一套无人机检测和定位的软硬件系统,试验结果表明该方法是可行、有效的,定位精度满足要求。In this paper,the author proposes a real-time unmanned aerial vehicle auto localization method based on multi-view stereo vision.Firstly,cameras are divided into several groups,and each group is connected with a computer to construct a local area network.Then,interior and exterior elements as well as the lens distortion parameters are measured.Next,images are captured simultaneously and the unmanned aerial vehicle is detected and located automatically by background difference and forward intersection. In order to verify this method,an unmanned aerial vehicle localization system is designed and implemented.The experimental results show that the method is feasible and effective and the expected accuracy is reached.

关 键 词：无人机自动定位 局域网组建 多目立体视觉 背景差分 

分 类 号：P237[天文地球—摄影测量与遥感]