摘要
针对单目视觉/INS的组合定位算法中,单目视觉本身不具备尺度的问题以及在景深变化明显和光照不均匀的环境下会产生粗差的问题,提出了一种单目视觉/INS组合室内定位抗差方法。在尺度恢复方面,借助INS的位置增量信息,基于最小二乘法对单目视觉进行尺度恢复。在抗差方法方面,利用INS在短时间间隔内增量信息高精度的特性,对单目视觉在此时间间隔内的增量信息进行检测,判断是否产生粗差并进行粗差剔除,进而提高组合定位系统的定位精度。实验结果表明,所提算法可以有效的探测并剔除单目视觉/INS组合系统在恶劣环境下产生的粗差,提高组合定位系统的精度及鲁棒性。
In the light of the problem of monocular vision/INS,monocular vision itself does not have scale,and it will produce gross errors in the environment of obvious change of depth of field and uneven illumination.A monocular vision/INS integrated indoor positioning robust method is proposed.In the aspect of scale restoration,with the help of INS position increment information,the scale restoration of monocular vision is based on least square method.In the aspect of robust method,the incremental information of monocular vision in this time interval is detected by using the high accuracy of INS incremental information in a short time interval to judge whether there are gross errors and eliminate them,so as to improve the positioning accuracy of the combined positioning system.The experimental results show that the proposed algorithm can effectively detect and eliminate gross errors caused by monocular vision/INS integrated system in harsh environments,and improve the accuracy and robustness of the integrated positioning system.
作者
徐爱功
宋帅
隋心
乔智
高嵩
XU Aigong;SONG Shuai;SUI Xin;QIAO Zhi;GAO Song(School of Geomatics,Liaoning Technical University,Fuxin,Liaoning 123000,China)
出处
《测绘科学》
CSCD
北大核心
2019年第12期1-6,共6页
Science of Surveying and Mapping
基金
国家重点研发计划项目（2016YFC0803102）
辽宁省教育厅辽宁省高等学校基本科研项目（LJ2017QL007）
关键词
尺度恢复
粗差剔除
惯性导航
视觉SLAM
组合定位
scale recovery
gross error removal
inertial navigation
visual SLAM
integrated positioning
