LTI-UAV3D: a solution of high quality and low cost for terrain 3D reconstruction using a μ-uav

被引:1
作者
Ramirez-Torres, Jose-Gabriel [1 ]
Larranaga-Cepeda, Ander [1 ]
机构
[1] CINVESTAV, Unidad Tamaulipas, Parque Cient & Tecnol TECNOTAM, Victoria 87130, Tamaulipas, Mexico
来源
DYNA | 2017年 / 92卷 / 04期
关键词
UAV; three-dimensional reconstruction; image processing; height map; real time;
D O I
10.6036/8174
中图分类号
T [工业技术];
学科分类号
08 ;
摘要
This article describes a simple and quite affordable tool called LTIUAV3D, for terrain three-dimensional reconstruction, using a low-cost unmanned aerial vehicle (UAV) and a personal computer. The 3D model does not require GPS information, because it is built exclusively from monocular images obtained by the onboard camera. To solve the scale factor of the 3D model, an artificial marker is used as a reference, so the obtained height map is metrically precise and correct. An important characteristic of LTIUAV3D is that the 3D model is built in real time, as the UAV flies over the terrain, with incremental updates every second. For the height map obtained, the estimation errors are less than 4%, so the 3D model could be used for path planning and autonomous navigation of UAVs. Also, the LTIUAV3D tool can get a much more precise and detailed 3D model, in an offline process, with height estimation errors less than 1%. This tool makes available, to a large number of users both individuals and academics, a robust foundation for the development of new applications of UAVs.
引用
收藏
页码:412 / 419
页数:8
相关论文
共 34 条
  • [1] [Anonymous], THESIS
  • [2] Barazzetti L., 2013, PHOTOGRAMM REC, V25, P2010
  • [3] Bills C, 2011, IEEE INT CONF ROBOT
  • [4] Bybee TC, 2015, P SPIE INT SOC OPTIC
  • [5] 3D free-form object recognition in range images using local surface patches
    Chen, Hui
    Bhanu, Bir
    [J]. PATTERN RECOGNITION LETTERS, 2007, 28 (10) : 1252 - 1262
  • [6] Davison AJ, 2003, NINTH IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION, VOLS I AND II, PROCEEDINGS, P1403
  • [7] Engel J., 2012, INTELLIGENT ROBOTS S, V320
  • [8] Accurate, Dense, and Robust Multiview Stereopsis
    Furukawa, Yasutaka
    Ponce, Jean
    [J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2010, 32 (08) : 1362 - 1376
  • [9] Automatic generation and detection of highly reliable fiducial markers under occlusion
    Garrido-Jurado, S.
    Munoz-Salinas, R.
    Madrid-Cuevas, F. J.
    Marin-Jimenez, M. J.
    [J]. PATTERN RECOGNITION, 2014, 47 (06) : 2280 - 2292
  • [10] Gertler J., 2012, LIBR C WASH DC CONGR