摘要
局部三维地图构建与移动轨迹跟踪在无人机导航、移动机器人和AR等众多领域有着非常重要的作用.本文采用改进型DAISY特征描述符对视频图像序列进行特征匹配跟踪,再利用对极约束求解基础矩阵,结合对极几何与射影几何关系得到相机参数与三维空间点坐标,最后利用增量BA优化三维稀疏地图和相机移动轨迹.实验表明本算法的特征点正确匹配率、三维点重建精度与完整度均有提高,能够输出清晰的相机移动轨迹和三维稀疏地图.
Local 3 D map construction and moving trajectory tracking play an important role in many fields,such as UAV navigation,mobile robot and AR.In this paper,the improved Daisy feature descriptor is used to match and track feature points,and then the epipolar constraint is used to solve the fundamental matrix.The epipolar geometry and projective geometry are combined to obtain the camera parameters and 3 D space point coordinates.Finally,the incremental BA is used to optimize the 3 D sparse map and camera moving trajectory.Experimental results show that the correct matching rate,the accuracy and completeness of 3 D point reconstruction are improved.The clear camera trajectory and 3 D sparse map can be obtained.
作者
李亚兰
曹江
陆汝华
黄健全
蒋纯志
Li Yalan;Cao Jiang;Lu Ruhua;Huang Jianquan;Jiang Chunzhi(School of Physics and Electronic Electrical Engineering,Xiangnan University,Chenzhou 423000,China;School of Computer and Artificial Intelligence,Xiangnan University,Chenzhou 423000,China)
出处
《湘南学院学报》
2022年第2期21-27,共7页
Journal of Xiangnan University
基金
湖南省社会科学成果评审委员会课题(XSP21YBC149)
湖南省大学生创新训练项目(S202110545032)
郴州市社会科学规划项目(czssk12021053,czssk12021071)
湘南学院应用特色学科项目(湘南学院校发[2018]108号No.6)
关键词
三维重建
移动定位
特征匹配
three dimensional reconstruction
mobile localization
feature matching
