单目相机姿态估计的点云与图像融合

发布时间：2018-07-08 10:22

  本文选题：姿态估计 + 时间匹配　； 参考：《测绘科学》2016年02期

【摘要】：针对点云与图像信息快速融合的问题,该文研究并实现了一种基于单目相机姿态估计的点云与图像融合方法。首先利用视觉导航中单目相机姿态估计方法估算小型无人机的姿态信息,然后与无人机导航系统中的姿态信息匹配来准确获得相机曝光时间点,最后结合图像曝光时间点的外方位元素和已知的内方位元素完成点云与图像融合。该方法直接利用图像信息进行融合,省去了相机时间检校步骤,避免了由此带来的误差,提高了时间匹配的准确性以及数据融合的自动化水平。实验结果表明,该方法能够准确地获得时间匹配信息,融合效果良好。
[Abstract]:Aiming at the problem of fast fusion of point cloud and image information, a method of point cloud and image fusion based on single camera attitude estimation is studied and implemented in this paper. Firstly, the attitude information of small UAV is estimated by using the attitude estimation method of monocular camera in visual navigation, and then the camera exposure time is accurately obtained by matching the attitude information with the attitude information in UAV navigation system. Finally, the point cloud is fused with the external azimuth element of the image exposure time point and the known interior azimuth element. This method directly uses image information to fuse, saves camera time check step, avoids the error caused by it, improves the accuracy of time matching and the automation level of data fusion. Experimental results show that this method can accurately obtain time matching information, and the fusion effect is good.
【作者单位】： 华南理工大学自动化科学与工程学院;
【基金】：国家重点基础研究发展计划(973计划)资助项目(2014CB845301/2/3) 2013年度开放基金项目(1214) 华南理工大学中央高校基本科研业务费自助项目(2014ZP0021)
【分类号】：P23
，

本文编号：2107121