一种基于惯性/高度计/视觉融合的地理信息修正算法

A georeferencing algorithm based on the inertial/altimeter/vision fusion

针对无人机视觉景象匹配导航中感知图像地理信息修正精度与实时性相互制约的问题,提出了一种基于惯性/高度计/视觉融合的地理信息修正算法。首先基于惯性导航解算的姿态与相机内参完成透视变换矩阵的计算,并通过线性插值完成实时图像的姿态修正,同时在该过程中引入透视变换矩阵因子降低传感器标定误差对修正精度的影响。进一步利用相对高度数据与相机内参完成空间分辨率计算及尺度修正。最终使其与基准图的坐标系、空间分辨率相同。为验证所设计算法的效果,设计了惯性/高度计/视觉组合装置并进行了飞行实验。实验结果表明地理信息修正平均误差为0.84像素,时间耗费为0.043 s。与传统利用图像匹配进行修正的算法相比,所提算法平均误差减小了80%,耗费时间缩短了90%以上,在精度与实时性上均有大幅提升。

In order to solve the problem that the geographic information correction of the real-time image cannot have both high precision and strong real-time performance in UAV visual scene matching navigation,a geographic information correction algorithm based on inertia/altimeter/vision fusion is proposed.Firstly,the perspective transformation matrix is calculated based on the attitude computed by inertial navigation and the internal parameters of the camera,and the attitude of the real-time image is corrected by linear interpolation.At the same time,perspective transformation matrix factor is introduced in the process to reduce the influence of sensor calibration error on the correction accuracy.Then,the relative height data and the internal parameters of the camera are further used to calculate the spatial resolution and modify the scale of the real-time image.Finally,the coordinate system and spatial resolution of the processed image are the same as that of the reference map.In order to verify the effectiveness of the proposed algorithm,a combined inertial/altimeter/vision device is designed and a flight experiment is carried out.The experimental results show that the average error of geographical information correction is 0.84 pixels,and the time consumption is 0.043 s.Compared with the traditional correction algorithm using image matching,the average error of the proposed algorithm is reduced by 80%,the time is shortened by more than 90%,and the accuracy and real-time performance are greatly improved.