基于约束Delaunay三角形的多视3D重建
|
摘要
为解决多视立体重建的表面模型扭曲变形,建筑物平面凹凸不平进而造成纹理贴图出现错误和拉花的问题,提出一种加入直线约束的城市建筑三维重建方法。该方法使用无人机获得的一系列图片作为输入,首先通过直线检测器和超级像素提取图像直线段和边缘,经边缘多边形简化,再通过2D约束德劳内三角化得到单视图2D三角格网。然后根据密集点云回算2D三角格网顶点三维位置,得到单视图3D格网模型,最后多视图合并得到场景完整3D模型。实验结果表明,算法表现出较好的性能、稳定性和可扩展性,并且具有较高精确性和有效性。
In order to solve the problem that the surface model of multi-view-stereo reconstruction is distorted and the plane of the building is uneven,which causes the error of texture mapping and garland,a method of 3D reconstruction of urban buildings with linear constraint is proposed. The method uses a series of pictures obtained by the UAV as input.Firstly,straight line segment and edge are extracted by line segment detector and super-pixel. After the polygon simplified,2D single triangle 2D mesh is obtained by 2D constrained Delaunay triangulation. Then,according to the dense point cloud,the 3D position of 2D triangular mesh vertex is back-calculated,and the single view 3D mesh model is obtained. Finally,the multi-view merges to obtain the complete scene 3D model. The experimental results show that the proposed algorithm has good performance,stability and expansibility,and has high accuracy and efficiency.
In order to solve the problem that the surface model of multi-view-stereo reconstruction is distorted and the plane of the building is uneven,which causes the error of texture mapping and garland,a method of 3D reconstruction of urban buildings with linear constraint is proposed. The method uses a series of pictures obtained by the UAV as input.Firstly,straight line segment and edge are extracted by line segment detector and super-pixel. After the polygon simplified,2D single triangle 2D mesh is obtained by 2D constrained Delaunay triangulation. Then,according to the dense point cloud,the 3D position of 2D triangular mesh vertex is back-calculated,and the single view 3D mesh model is obtained. Finally,the multi-view merges to obtain the complete scene 3D model. The experimental results show that the proposed algorithm has good performance,stability and expansibility,and has high accuracy and efficiency.
引文
[1]Seitz S M,Curless B,Diebel J,et al.A comparison and evaluation of multi-view stereo reconstruction algorithms[C]//Computer vision and pattern recognition,2006 IEEE Computer Society Conference on.IEEE,2006,1:519-528.
[2]Micˇu2ík B,Ko2eckáJ.Multi-view Superpixel Stereo in Urban Environments[J].International Journal of Computer Vision,2010,89(1):106-119.
[3]欧元汉,黄鸿,陆宇海.基于密集点云重投影的三维线段模型重建算法[J].计算机工程与设计,2015(12):3291-3296.
[4]罗宇,宋歌,杨红雨.基于图像识别的建筑物三维重建[J].计算机工程与设计,2015,36(1):191-196.
[5]Hofer M,Wendel A,Bischof H.Line-based 3D reconstruction of wiry objects[C]//Computer Vision Winter Workshop(CVWW),2013.
[6]Qi M,Cao T T,Tan T S.Computing 2D constrained Delaunay triangulation using the GPU[J].Visualization and Computer Graphics,IEEE Transactions on,2013,19(5):736-748.
[7]Furukawa Y,Ponce J.Accurate,dense,and robust multiview stereopsis[J].Pattern Analysis and Machine Intelligence,IEEE Transactions on,2010,32(8):1362-1376.
[8]Furukawa Y,Curless B,Seitz S M,et al.Reconstructing building interiors from images[C]//Computer Vision,2009IEEE 12th International Conference on.IEEE,2009:80-87.
[9]Gallup D,Frahm J M,Mordohai P,et al.Real-time planesweeping stereo with multiple sweeping directions[C]//Computer Vision and Pattern Recognition,2007.CVPR’07.IEEE Conference on.IEEE,2007:1-8.
[10]Vu H H,Labatut P,Pons J P,et al.High accuracy and visibility-consistent dense multiview stereo[J].Pattern Analysis and Machine Intelligence,IEEE Transactions on,2012,34(5):889-901.
[11]Sugiura T,Torii A,Okutomi M.3D surface reconstruction from point-and-line cloud[C]//3D Vision(3DV),2015 International Conference on.IEEE,2015:264-272.
[12]Bódis-SzomorúA,Riemenschneider H,Van Gool L.Superpixel meshes for fast edge-preserving surface reconstruction[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2015:2011-2020.
[13]Gioi R G V,Jakubowicz J,Morel J M,et al.LSD:A line segment detector[J].Image Processing on Line,2012,2(4):35-55.
[14]Ren C Y,Reid I.g SLIC:a real-time implementation of SLICsuperpixel segmentation[R].University of Oxford,Department of Engineering,Technical Report,2011.
[15]Park J H,Jang B J,Kwon O J,et al.Vector Map Data compression based on Douglas Peucker Simplification Algorithm and Bin Classification[J].Journal of Korea Multimedia Society,2015,18(3):298-311.
[16]Jancosek M,Pajdla T.Exploiting visibility information in surface reconstruction to preserve weakly supported surfaces[J].International Scholarly Research Notices,2014,2014:1-20.
[17]Sinha S N,Mordohai P,Pollefeys M.Multi-view stereo via graph cuts on the dual of an adaptive tetrahedral mesh[C]//Computer Vision,2007.ICCV 2007.IEEE 11th International Conference on.IEEE,2007:1-8.
[18]Wu C.Visual SFM:A visual structure from motion system[Z].2011.
[19]Li Z,Wang K,Zuo W,et al.Detail-preserving and Content-aware Variational Multi-view Stereo Reconstruction[J].Image Processing,IEEE Transactions on,2016,25(2):864-877.
[2]Micˇu2ík B,Ko2eckáJ.Multi-view Superpixel Stereo in Urban Environments[J].International Journal of Computer Vision,2010,89(1):106-119.
[3]欧元汉,黄鸿,陆宇海.基于密集点云重投影的三维线段模型重建算法[J].计算机工程与设计,2015(12):3291-3296.
[4]罗宇,宋歌,杨红雨.基于图像识别的建筑物三维重建[J].计算机工程与设计,2015,36(1):191-196.
[5]Hofer M,Wendel A,Bischof H.Line-based 3D reconstruction of wiry objects[C]//Computer Vision Winter Workshop(CVWW),2013.
[6]Qi M,Cao T T,Tan T S.Computing 2D constrained Delaunay triangulation using the GPU[J].Visualization and Computer Graphics,IEEE Transactions on,2013,19(5):736-748.
[7]Furukawa Y,Ponce J.Accurate,dense,and robust multiview stereopsis[J].Pattern Analysis and Machine Intelligence,IEEE Transactions on,2010,32(8):1362-1376.
[8]Furukawa Y,Curless B,Seitz S M,et al.Reconstructing building interiors from images[C]//Computer Vision,2009IEEE 12th International Conference on.IEEE,2009:80-87.
[9]Gallup D,Frahm J M,Mordohai P,et al.Real-time planesweeping stereo with multiple sweeping directions[C]//Computer Vision and Pattern Recognition,2007.CVPR’07.IEEE Conference on.IEEE,2007:1-8.
[10]Vu H H,Labatut P,Pons J P,et al.High accuracy and visibility-consistent dense multiview stereo[J].Pattern Analysis and Machine Intelligence,IEEE Transactions on,2012,34(5):889-901.
[11]Sugiura T,Torii A,Okutomi M.3D surface reconstruction from point-and-line cloud[C]//3D Vision(3DV),2015 International Conference on.IEEE,2015:264-272.
[12]Bódis-SzomorúA,Riemenschneider H,Van Gool L.Superpixel meshes for fast edge-preserving surface reconstruction[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2015:2011-2020.
[13]Gioi R G V,Jakubowicz J,Morel J M,et al.LSD:A line segment detector[J].Image Processing on Line,2012,2(4):35-55.
[14]Ren C Y,Reid I.g SLIC:a real-time implementation of SLICsuperpixel segmentation[R].University of Oxford,Department of Engineering,Technical Report,2011.
[15]Park J H,Jang B J,Kwon O J,et al.Vector Map Data compression based on Douglas Peucker Simplification Algorithm and Bin Classification[J].Journal of Korea Multimedia Society,2015,18(3):298-311.
[16]Jancosek M,Pajdla T.Exploiting visibility information in surface reconstruction to preserve weakly supported surfaces[J].International Scholarly Research Notices,2014,2014:1-20.
[17]Sinha S N,Mordohai P,Pollefeys M.Multi-view stereo via graph cuts on the dual of an adaptive tetrahedral mesh[C]//Computer Vision,2007.ICCV 2007.IEEE 11th International Conference on.IEEE,2007:1-8.
[18]Wu C.Visual SFM:A visual structure from motion system[Z].2011.
[19]Li Z,Wang K,Zuo W,et al.Detail-preserving and Content-aware Variational Multi-view Stereo Reconstruction[J].Image Processing,IEEE Transactions on,2016,25(2):864-877.