无人机航拍图像的三维重建方法研究  

Research on 3D Reconstruction Method of Aerial Image of Drone

作　　者：周祖鹏[1] 张晓东 甘良棋 ZHOU Zu-peng;ZHANG Xiao-dong;GAN Liang-qi(College of Mechanical and Electrical Engineering,Guilin University of Electronic Technology,Guangxi Guilin 541000,China)

机构地区：[1]桂林电子科技大学机电工程学院,广西桂林541000

出　　处：《机械设计与制造》2021年第5期47-50,55,共5页Machinery Design & Manufacture

基　　金：带机械臂的四旋翼无人机协同控制方法研究(61763006)。

摘　　要：在无人机为载体的基础上进行对目标场景的三维重建,就是结合无人机和计算机视觉技术,利用无人机操作灵活性,视角可控制性等优点。为实现更加完整的三维模型的重建,提出一种基于多图像拼接三维重建算法。基本思路是在无人机为载体的基础上,从不同方向获取目标物体的图像,通过自标定方法获取相机内参数,采用图像拼接融合技术对多幅图像分析、合成,从而最大限度地对建筑物场景的各种特征信息的描述。进一步对拼接融合后的图像进行特征点提取和点云匹配,从而获取全景图空间特征点三维点云,获得一个较为真实的重构对象的三维模型。实验结果表明,改进后的重构方法的精度较高,适合在许多场景三维重建的应用。On the basis of the UAV as the carrier,the three-dimensional reconstruction of the target scene is combined with the UAV and computer vision technology,utilizing the operational flexibility of the UAV and the controllability of the viewing angle.In order to realize the reconstruction of a more complete 3D model,a multi-image mosaic 3D reconstruction algorithm is proposed.The basic idea is to obtain the image of the target object from different directions on the basis of the UAV as the carrier,obtain the parameters of the camera by self-calibration method,and analyze and synthesize multiple images by image mosaic fusion technology to maximize the construction of the camera.Description of various feature information of the scene.The feature points extraction and point cloud matching are further performed on the stitched and merged images,so as to obtain a three-dimensional point cloud of the panoramic feature points of the panoramic image,and obtain a realistic three-dimensional model of the reconstructed object.The experimental results show that the improved reconstruction method has higher precision and is suitable for 3D reconstruction in many scenes.

关 键 词：无人机 自标定 图像拼接 匹配 三维重建 

分 类 号：TH16[机械工程—机械制造及自动化] TP391.9[自动化与计算机技术—计算机应用技术]