Surface Reconstruction of Point Cloud Based on Geometric Structure

doi:10.16182/j.issn1004731x.joss.201711013

Abstract

Abstract: To solve the problems that the existing noise in dense point cloud generated from sequential images dense matching affects the accuracy, and geometric structure such as flat exists in urban scenes, a surface reconstruction method of dense point cloud based on geometric structure is proposed in this paper. RANSAC algorithm is applied to extract planar structure. The original points were structured into planar points, crease points of two planes intersecting, corner points of three or more planes intersecting and clutter points. Then the structured point cloud is divided into tetrahedron by 3D Delaunay, improving the penalty term of min-cut energy function by the triangular patches consisted by the four structured points to extract surface of the 3D Delaunay tetrahedron. The experiment results show that, compared with several classical methods and business software, the proposed method can reconstruct a real 3D scene and can recovery the scene geometry structure of plane.

Key words: sequential images, dense point cloud, surface reconstruction, 3D Delaunay partition, min-cut

CLC Number:

TP274

Yang Zhenfa, Wan Gang, Cao Xuefeng, Li feng, Xie Lixiang. Surface Reconstruction of Point Cloud Based on Geometric Structure[J]. Journal of System Simulation, 2017, 29(11): 2684-2692.

References

[1] 刘春, 陈华云, 吴杭彬. 激光三维遥感的数据处理与特征提取[M]. 北京: 科学出版社, 2009.
LIU Chun, CHEN Huayun, WU Hangbin.Data Processing and Feature Extraction of Laser 3D Remote Sensing [M]. Beijing: Science Press, 2009.
[2] 马颂德,张正友. 计算机视觉——计算理论与算法基础[M] 北京: 科学出版社, 1998.
MA Songde, ZHANG Zhengyou.Computer vision theory and algorithms [M] Beijing: Science Press, 1998.
[3] Edelsbrunner H, Mucke E P.Three-dimentional Alpha Shapes[J]. ACM Transaction on Graphics (S0730-0301), 1994, 13(1): 43-72.
[4] Amenta N, Bern M, Kamvysselisy M.A new Voronoi- based surface reconstruction algorithm[C]//ACM SIGGRAPH (1998), 1998: 415-421.
[5] Amenta N, Choi S, Dey T K, et al.Asimple algorithm for homeomorphic surface reconstruction[C]// International Journal of Computational Geometry and Applications. New York: ACM Press, 2000: 213-222.
[6] V H Hiep, R Keriven, P Labatut, et al.Towards high-resolution large-scale multi-view stereo[C]// Computer Vision and Pattern Recognition, 2009.
[7] Michael Kazhdan, Matthew Bolitho, Hugues Hoppe.Poisson Surface Reconstruction[C]//the forth Eurographics symposium on Geometry, Switzerland, 2006.
[8] Florent Lafarge, Clément Mallet.Creating large-scale city models from 3d-point clouds: a robust approach with hybrid representation[J]. International Journal of Computer Vision (S0920-5691), 2012, 99(1):69-85.
[9] Furukawa Y, Curless B, Seitz S M, et al.Manhattan- world stereo[C]// CVPR , IEEE Computer Society, 2009.
[10] Snavely N, Seitz S, Szeliski R.Photo tourism: exploring photo collections in 3D[J]. ACM Transaction on Graphics (S0730-0301), 2006, 25(3): 835-846.
[11] Yasutaka Furukawa, Jean Ponce. Accurate, Dense,Robust Multi-View Stereopsis[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence (S0162-8828), 2010, 32(8): 1362-1376.
[12] Changchang Wu, Sameer Agarwal, Brian Curless, et al.Multicore bundle adjustment[C]// IEEE Computer Vision and Pattern Recognition, 2011.
[13] 魏英姿, 刘晓莉. 基于随机抽取一致性的稳健点云平面拟合[J]. 北京工业大学学报, 2014, 40(3): 400-403.
WEI Yingzi, LIU Xiaoli.Robust Plane Fitting of Point Cloud Based on RANSAC[J]. Journal of Beijing University of Technology. 2014, 40(3): 400-403.
[14] Tsai V J D. Delaunay triangulation in TIN creation: an oveview and linear time algorithm[J]. International Journal of Geographical Information Science (S1365-8816), 1993, 7(6): 501-524.
[15] D Attli, J D Boissonnat, A Lieutier.Complexity of the Delaunay triangulation of points on surfaces: the smooth case[C]// Nineteenth Symposium on Computational Geometry. ACM, 2003:201-210.
[16] Labatut P, Pons J P, Keriven R.Efficient multi-view reconstruction of large-scale scenes using interest points, Delaunay triangulation and graph cuts[C]//IEEE International Conference on Computer Vision, Rio de Janeiro, 2007.
[17] Labatut P, Pons J ‐, Keriven R. Robust and Efficient Surface Reconstruction From Range Data[J]. Computer Graphics Forum (S1467-8659), 2010, 28(8): 2275-2290.
[18] Pix4d. Pix4D Support Site[EB/OL].https: // support. pix4d. com/hc/ en-us/ sections/ 200591139- Example- Datasets- Available- for- Download # gsc. tab= 0, 2015.
[19] CVLab. CVLab Data[EB/OL]. http://cvlabwww. Epfl. ch/ data/ multiview/ denseMVS. Html, 2015.
[20] Boykov Y, Kolmogorov. An experimental comparison of min-cut/max-flow algorithms for energy minimization in vison[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence (S0162-8828), 2004, 26(9): 1124-1137.
[21] Chen Zitan, Guevara J A.Systematic Selection of Very ImPORTANT Points from Digital Terrain Model for Constructing Triangular Irregular Net-Works[C]// Auto-Carto 8, Baltimore, USA, 1987.
[22] M P Dubuisson, A K Jain.A modified Hausdorff distance for object matching[C]//The 12th IAPR International Conference, Jerusalem, 1994.