二维卫星地图像素匹配的无人机视惯定位方法

In order to improve the location accuracy of the unmanned aerial vehicle (UAV) visual inertial system, a visual inertial location method using Internet 2D satellite maps and UAV aerial images matching is proposed. Firstly, the aerial images and IMU are used to construct the visual inertial odometry (VIO) to estimate the UAV motion state and obtain the depth of feature points. Then extract feature points from the matched image pairs, perform pixel-level matching and solving based on the known satellite map scale and the depth of feature points estimated by VIO, and obtain the matching and location results. Finally, the VIO factor and the matching location factor are constructed, and the factor map is used to optimize the global position. The proposed method is validated by airborne experiments, and the experimental results show that the average location error is reduced by 50.14% compared with the VIO location method without map matching, and the average location error is reduced by 43.69% compared with the VIO and loopback location method without map matching. The location accuracy of the visual inertial system is effectively improved.