由非专业弱关联影像自动化构建的三维地理空间模型是地理空间信息的重要来源。非专业弱关联影像在三维重建后必须经过地理配准,具有了绝对地理空间坐标系的位置信息及其准确的空间精度信息后,才有可能成为有效的地理空间信息。本文提出了一种以相机GPS模块获取的地理空间坐标为依据的理配准方法,依据影像的地理空间坐标和其三维重建后得到图像空间坐标的空间相似性,考虑GPS实时测量坐标精度较差和高程测量值不稳定的特点,采用RANSAC方法求解二维和三维两种空间变换参数及地理配准结果。利用差分GPS测量的影像位置数据对地理配准的精度进行了分析,给出了位移、旋转和缩放等误差的定量评估结果,分析了产生错误结果的原因。这种地理配准方法对数据采集设备要求低,过程无须人工参与。试验证明,在参与地理配准运算的照片数量较多时,配准结果正确、空间精度较高。
The 3D geo-spatial model built by unprofessional and weakly-related image is a significant source of geo-spatial information. The unprofessional and weakly-related image cannot be useful geo-spatial information until be geo-registered with accurate geo-spatial orientation and location. In this paper, we present an automatic geo-registration using the coordination acquired by real-time GPS module. We calculate 2D and 3D spatial transformation parameters based on the spatial similarity between the image location in the geo-spatial coordination system and in the 3D reconstruction coordination system. Because of the poor precision of GPS information and especially the unstability of elevation measurement, we use RANSAC algorithm to get rid of outliers. In the experiment, we compare the geo-registered image positions to their differential GPS coordinates. The errors of translation, rotation and scaling are evaluated quantitively and the causes of bad result are analyzed. The experiment demonstrates that this geo-registration method can get a precise result with enough images.
[1] LIU Yingzhen, JIA Fenli, WAN Gang, et al. Construction and Application of 3D GIS Based on Unprofessionaland Weakly-Correlated Image[J]. Journal of Geomatics Science and Technology, 2014, 31(1): 73-78. (刘颖真, 贾奋励, 万刚, 等. 非专业弱关联影像构建三维GIS研究[J]. 测绘科学技术学报, 2014, 31(1): 73-78.)
[2] KAMINSKY R S, SNAVELY N, SEITZ S M, et al. Alignment of 3D Point Clouds to Overhead Images[C]//IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops. Miami: IEEE, 2009:63-70.
[3] WANG Chunpo, WILSON K, SNAVELY N. Accurate Georegistration of Point Clouds Using Geographic Data[C]//2013 International Conference on 3D Vision. Seattle: IEEE, 2013:33-40.
[4] ROBERTSON D P, CIPOLLA R. Building Architectural Models from Many Views Using Map Constraints[C]//HEYDEN A,SPARR G,NIELSEN M, et al.The European Conference on Computer Vision 2002. Copenhagen: Springer, 2002: 155-169.
[5] CHO P. 3D Organization of 2D Urban Imagery[C]//Applied Imagery Pattern Recognition Workshop. Washington: IEEE, 2007:3-8.
[6] CHO P, SNAVELY N. Enhancing Large Urban Photo Collections with 3D LiDAR and GIS Data[J]. International Journal of Remote Sensing Applications, 2013, 3(1): 1-10.
[7] CHO P, SNAVELY N. 3D Exploitation of 2D Ground-level & Aerial Imagery[C]//IEEE Applied Imagery Pattern Recognition Workshop. Washington, DC: IEEE, 2011:1-8.
[8] NI K, SUN Z, BLISS N. 3D Image Geo-Registration Using Vision-based Modeling[C]//IEEE International Conference on Acoustics, Speech and Signal Processing. Prague: IEEE, 2011:1573-1576.
[9] WENDEL A, IRSCHARA A, BISCHOF H. Automatic Alignment of 3D Reconstructions Using a Digital Surface Model[C]//IEEE Computer Society Computer Vision and Pattern Recognition Workshops. Colorado Springs: IEEE, 2011:29-36.
[10] WENDEL A, BISCHOF H.Visual Localization for Micro Aerial Vehicles in Urban Outdoor Environments[M]//FARINELLA G M, BATTIATO S,CIPOLLA R. Advanced Topics in Computer Vision. London: Springer, 2013: 181-214.
[11] WENDEL A, MAURER M, BISCHOF H. Visual Landmark-based Localization for MAVs Using Incremental Feature Updates[C]//2012 Second International Conference on 3D Imaging, Modeling, Processing, Visualization & Transmission. Zurich: IEEE, 2012:278-285.
[12] WENDEL A, IRSCHARA A, BISCHOF H. Natural Landmark-based Monocular Localization for MAVs[C]//2011 IEEE International Conference on Robotics and Automation. Shanghai: IEEE, 2011:5792-5799.
[13] SHEN Yonglin, LIU Jun, WU Lixin, et al. Reconstruction of Disaster Scene from UAV Images and Flight-control Data[J]. Geography and Geo-Information Science, 2011, 27(6): 13-17. (沈永林, 刘军, 吴立新, 等. 基于无人机影像和飞控数据的灾场重建方法研究[J]. 地理与地理信息科学, 2011, 27(6): 13-17.)
[14] FRAHM J M, HEINLY J, ZHENG Enliang, et al. Geo-Registered 3D Models from Crowdsourced Image Collections[J]. Geo-spatial Information Science, 2013, 16(1): 55-60.
[15] ZHANG Liang, MA Hongchao, GAO Guang, et al. Automatic Registration of Urban Aerial Images with Airborne LiDAR Points Based on Line-point Similarity Invariants[J].Acta Geodaetica et Cartographica Sinica, 2014, 43(4): 372-379. (张良, 马洪超, 高广, 等. 点、线相似不变性的城区航空影像与机载激光雷达点云自动配准[J]. 测绘学报, 2014, 43(4): 372-379.)
[16] LI Tianwen. Theory and Application of GPS[M]. Beijing: Science Press, 2003:92-93. (李天文. GPS原理及应用[M]. 北京: 科学出版社, 2003:92-93.)
[17] JIA Yunde. Machine Vision[M]. Beijing: Science Press, 2000:191-192. (贾云得. 机器视觉[M]. 北京: 科学出版社, 2000:191-192.)
[18] SZELISKI R. Image Alignment and Stitching: A Tutorial[J]. Foundations and Trends in Computer Graphics and Vision, 2006, 2(1): 1-104.
[19] FISCHLER M A, BOLLES R C. Random Sample Consensus: A Paradigm for Model Fitting with Applications to Image Analysis and Automated Cartography[J]. Communications of the ACM, 1981, 24(6): 381-395.
[20] NIKON. GPS Unit GP-1[EB/OL].[2014-06-28]. http://imaging.nikon.com/lineup/accessory/camera/gp-1/spec.htm.
[21] SOUTH GROUP. RTK Surveying System >>S82T[EB/OL].[2014-07-08]. http://www.southsurvey.com/public/xianxi.php?id=301. (南方测绘. RTK测量系统>>S82T[EB/OL].[2014]. http://www.southsurvey.com/public/xianxi.php?id=301.)
