Computer Engineering and Applications ›› 2018, Vol. 54 ›› Issue (1): 217-223.DOI: 10.3778/j.issn.1002-8331.1607-0155

Previous Articles     Next Articles

Dense 3D reconstruction based on removing camera-shake fuzzy algorithm

ZHENG En, CHENG Yaotian, LIN Jingyu   

  1. College of Electrical Engineering, Guangxi University, Nanning 530004, China
  • Online:2018-01-01 Published:2018-01-15


郑  恩,成耀天,林靖宇   

  1. 广西大学 电气工程学院,南宁 530004

Abstract: Some issues are prominent and unsettled in the field of Unmanned Aerial Vehicle(UAV) for big scene objects 3D reconstruction, for instance, dithering fuzzy phenomenon can still be attributed to camera-shake in 3D reconstruction; sparse point cloud and poor visualization are still in the presence after Structure from Motion(SFM) on 2D image sequences. To cope with the problems, this paper, at first, proposes the Removing Camera-Shake Fuzzy Algorithm to restore the original image information of the blurred image and then 3D reconstruction based on point cloud is put forward on the basis of structure from motion. Finally, Poisson surface reconstruction is applied into the point cloud obtained from the dense reconstruct. The experimental results demonstrate that the Removing Camera-shake Fuzzy Algorithm, to a large extent, improves the image quality and after dense 3D reconstruction, the big scene objects are more authentic and possess higher visual quality.

Key words: Unmanned Aerial Vehicle(UAV), shake the fuzzy, Structure from Motion(SFM), point cloud, dense reconstruction

摘要: 针对无人机在航拍大场景对象进行三维重建时因抖动产生的图像模糊现象,以及二维图像序列经运动恢复结构SFM后得到的点云较为稀疏,可视化差等不足,采用去抖动模糊算法恢复模糊图像的原始图像信息,然后在运动恢复结构的基础上进行基于点云的稠密三维重建,最后对稠密重建后的点云进行泊松表面重建以得到表面致密、均匀的三维模型。实验结果表明,去抖动模糊算法可以有效地提高图像的质量,大场景对象经过基于点云的稠密三维重建后得到的重建效果逼真,可视化强。

关键词: 无人机, 抖动模糊, 运动恢复结构, 点云, 稠密重建