摘要
分析了CCD相机所拍摄的序列影像中的特征点观测量,与惯性导航系统进行组合,提出了基于多状态约束的视觉/惯性组合导航算法,实现了在无GNSS环境下利用低成本惯性器件的导航定位。该算法可充分利用多帧连续影像中相同特征点所构建的几何约束关系,同时大大降低了计算量。通过仿真验证,组合导航算法的精度优于2m。
The point features in the sequence images captured by CCD camera are analyzed to integrate with the inertial navigation system.A multi-state constraint algorithm is established for vision-aided inertial navigation system,which realize navigation and positioning with low-cost inertial sensors in the GNSS denied environment.The proposed algorithm can make full use of the geometric constraints of the same point feature in multi-image frames.It also greatly reduces the computation amount.Verified by simulation,the accuracy of the integrated algorithm is better than 2meters.
出处
《光学与光电技术》
2015年第6期58-62,共5页
Optics & Optoelectronic Technology
关键词
组合导航
计算机视觉
卡尔曼滤波
多状态约束
integrated navigation
computer vision
Kalman filter
multi-state constraint
