一种基于无人机影像的高效三维重建方法

An Efficient 3D Reconstruction Method Based on UAV Images

为提升无人机影像三维重建的效率和稳健性,本文首先将定位测姿系统所获取的POS数据应用于大规模无人机影像三维重建当中,通过构建影像关系图,大大减少了影像匹配量,降低了误匹配率;其次,应用运动恢复结构技术中的李代数旋转平均方法解决了旋转参数的线性化求解问题,并在迭代解算过程提出了将L1范数与最小二乘相结合的计算方法,提高了模型解算成功率。实验结果表明:本文方法在多种条件下均能获取纹理清晰的地表重建模型,且处理效率得到了明显提升,可满足地理信息成果应急保障要求。

In order to improve the efficiency and robustness of UAV image 3D reconstruction,this paper first applies the POS data obtained by the positioning and attitude measurement system to the large-scale UAV image 3D reconstruction.By constructing the image relation graph,the amount of image matching and the rate of false matching are greatly reduced.Then,the linearization of rotation parameters is solved by using the Lie algebraic rotation average method in motion restoration structure technology,and a calculation method combining L1 norm and least square is proposed in the iterative solution process,which improves the success rate of model solution.The experimental results show that this method can obtain the surface reconstruction model with clear texture under various conditions,and the processing efficiency has been significantly improved to meet the requirements of emergency support of geographic information results.