基于深度学习的几何特征匹配方法
|
摘要
Kinect等深度相机采集的三维数据往往存在噪音、低分辨率等问题,导致两帧点云的局部几何特征匹配一直面临挑战。目前多采用基于特征直方图的方法解决这一问题,但其计算量较大,且对场景旋转平移的要求较为严格。文中提出了一种基于数据驱动的方法,首先从大量重建好的RGB-D数据集中,通过自监督的深度学习方法构建能够描述三维数据几何特征的模型;然后利用基于KD-Tree的K近邻算法(KNN)得到两部分点云的特征对应点,通过RANSAC剔除误匹配点对;最后通过得到的较准确的位置关系估计两帧点云的几何变换,从而完成配准。基于斯坦福大学点云库中的模型以及真实环境下Kinect采集到的大卫石膏像模型的配准和比较实验表明,所提方法不仅可以提取未知物体的局部几何特征进行配准,还可以较好地应对空间角度变换大的情况。
Matching local geometric features on real-world depth images is a challenging task due to the noisy and low-resolution of 3 D scan captured by depth cameras like Kinect.At present,most of the solutions to this problem are based on the feature histogram method,which requires a large amount of calculation and strict requirements on the rotation of the scene.This paper proposed a method based on data-driven.From a large number of well-reconstructed RGB-D data sets,a self-supervised deep learning method is used to construct a model that can describe the geometric correspondence between three-dimensional data.Then,corresponding approximate points of two parts of the point cloud are botained by using KD-Tree-based K Nearest Neighbor(KNN) algorithm.Through removing erroneous matching point pairs using RANSAC,a relatively accurate set of feature point pairs is obtained by estimating the geometric transformation.By regis-tering and comparing the models in the Stanford University point cloud library and the David plaster model collected in the real environment,the experiments show that the proposed method can not only extract the local geometric features of unknown objects for registration,but also show good performance when dealing with large changes in spatial angle.
Matching local geometric features on real-world depth images is a challenging task due to the noisy and low-resolution of 3 D scan captured by depth cameras like Kinect.At present,most of the solutions to this problem are based on the feature histogram method,which requires a large amount of calculation and strict requirements on the rotation of the scene.This paper proposed a method based on data-driven.From a large number of well-reconstructed RGB-D data sets,a self-supervised deep learning method is used to construct a model that can describe the geometric correspondence between three-dimensional data.Then,corresponding approximate points of two parts of the point cloud are botained by using KD-Tree-based K Nearest Neighbor(KNN) algorithm.Through removing erroneous matching point pairs using RANSAC,a relatively accurate set of feature point pairs is obtained by estimating the geometric transformation.By regis-tering and comparing the models in the Stanford University point cloud library and the David plaster model collected in the real environment,the experiments show that the proposed method can not only extract the local geometric features of unknown objects for registration,but also show good performance when dealing with large changes in spatial angle.
引文
[1] RUSU R B,BLODOW N,BEETZ M.Fast point feature histograms (FPFH) for 3D registration[C]//IEEE International Conference on Robotics and Automation.Kobe,Japan:IEEE Press,2009:3212-3217.
[2] MA D H,LIU G Z.Improved Method of Point Cloud Registration Based on FPFH Feature[J].Computer and Modernization,2017(11):46-50.(in Chinese)马大贺,刘国柱.改进的基于FPFH特征配准点云的方法[J].计算机与现代化,2017(11):46-50.
[3] BESL P J.A method for registration 3D shapes[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1992,14(2):193-200.
[4] ZHAO F Q,ZHOU M Q.Improved Probability Iterative Closest Point Registration Algorithm [J].Journal of Graphics,2017,38(1):15-22.(in Chinese)赵夫群,周明全.改进的概率迭代最近点配准算法[J].图学学报,2017,38(1):15-22.
[5] AIGER D,MITRA N J,COHEN-OR D.4-points congruent sets for robust pairwise surface registration[J].Acm Transactions on Graphics,2008,27(3):1-10.
[6] YU W L,ZHOU M Q,SHUI W Y,et al.Automatic Registration Method Based on Curvature[J].Journal of System Simulation,2015,27(10):2374-2379.(in Chinese)余文利,周明全,税午阳,等.基于曲率的点云自动配准方法[J].系统仿真学报,2015,27(10):2374-2379.
[7] MELLADO N,AIGER D,MITRA N J.Super 4PCS Fast Global Pointcloud Registration via Smart Indexing[J].Computer Graphics Forum,2015,33(5):205-215.
[8] WU Z,SONG S,KHOSLA A,et al.3D ShapeNets:A deep representation for volumetric shapes[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Boston,MA,USA,2014.
[9] ZENG A,SONG S,NIEBNER M,et al.3DMatch:Learning Local Geometric Descriptors from RGB-D Reconstructions[C]//IEEE Conference on Computer Vision and Pattern Recognition.IEEE Computer Society,2017:199-208.
[10] GUO K,ZOU D,CHEN X.3D Mesh Labeling via Deep Convolutional Neural Networks[J].Acm Transactions on Graphics,2015,35(1):1-12.
[11] ANGELO L D,GIACCARI L.An efficient algorithm for the nearest neighbourhood search for point clouds[J].International Journal of Computer Science Issues,2011,8(5):1-11.
[12] LEBEDA K,MATAS J,CHUM O.Fixing Locally Optimized RANSAC Full Experimental Evaluation[C]//British Machine Vision Conference.Guildford,UK,2012:1-11.
[13] KRIZHEVSKY A,SUTSKEVER I,HINTON G E.ImageNet classification with deep convolutional neural networks[C]//International Conference on Neural Information Processing Systems.Curran Associates Inc.2012:1097-1105.
[14] VALENTIN J,DAI A,NIEβNER M,et al.Learning to Navigate the Energy Landscape[C]//IEEE 2016 Fourth InternationalConference on 3D Vision (3DV).Stanford,CA,USA,2016:323-332.
[15] SHOTTON J,GLOCKER B,ZACH C,et al.Scene Coordinate Regression Forests for Camera Relocalization in RGB-D Images[C]//IEEE Conference on Computer Vision and Pattern Recognition.IEEE Computer Society,2013:2930-2937.
[16] XIAO J,OWENS A,TORRALBA A.SUN3D:A Database of Big Spaces Reconstructed Using SfM and Object Labels[C]//IEEE International Conference on Computer Vision.IEEE Computer Society,2013:1625-1632.
[17] The Stanford 3D Scanning Repository[EB/OL].(2014-08-19).www.graphics.st-anford.edu.
[2] MA D H,LIU G Z.Improved Method of Point Cloud Registration Based on FPFH Feature[J].Computer and Modernization,2017(11):46-50.(in Chinese)马大贺,刘国柱.改进的基于FPFH特征配准点云的方法[J].计算机与现代化,2017(11):46-50.
[3] BESL P J.A method for registration 3D shapes[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1992,14(2):193-200.
[4] ZHAO F Q,ZHOU M Q.Improved Probability Iterative Closest Point Registration Algorithm [J].Journal of Graphics,2017,38(1):15-22.(in Chinese)赵夫群,周明全.改进的概率迭代最近点配准算法[J].图学学报,2017,38(1):15-22.
[5] AIGER D,MITRA N J,COHEN-OR D.4-points congruent sets for robust pairwise surface registration[J].Acm Transactions on Graphics,2008,27(3):1-10.
[6] YU W L,ZHOU M Q,SHUI W Y,et al.Automatic Registration Method Based on Curvature[J].Journal of System Simulation,2015,27(10):2374-2379.(in Chinese)余文利,周明全,税午阳,等.基于曲率的点云自动配准方法[J].系统仿真学报,2015,27(10):2374-2379.
[7] MELLADO N,AIGER D,MITRA N J.Super 4PCS Fast Global Pointcloud Registration via Smart Indexing[J].Computer Graphics Forum,2015,33(5):205-215.
[8] WU Z,SONG S,KHOSLA A,et al.3D ShapeNets:A deep representation for volumetric shapes[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Boston,MA,USA,2014.
[9] ZENG A,SONG S,NIEBNER M,et al.3DMatch:Learning Local Geometric Descriptors from RGB-D Reconstructions[C]//IEEE Conference on Computer Vision and Pattern Recognition.IEEE Computer Society,2017:199-208.
[10] GUO K,ZOU D,CHEN X.3D Mesh Labeling via Deep Convolutional Neural Networks[J].Acm Transactions on Graphics,2015,35(1):1-12.
[11] ANGELO L D,GIACCARI L.An efficient algorithm for the nearest neighbourhood search for point clouds[J].International Journal of Computer Science Issues,2011,8(5):1-11.
[12] LEBEDA K,MATAS J,CHUM O.Fixing Locally Optimized RANSAC Full Experimental Evaluation[C]//British Machine Vision Conference.Guildford,UK,2012:1-11.
[13] KRIZHEVSKY A,SUTSKEVER I,HINTON G E.ImageNet classification with deep convolutional neural networks[C]//International Conference on Neural Information Processing Systems.Curran Associates Inc.2012:1097-1105.
[14] VALENTIN J,DAI A,NIEβNER M,et al.Learning to Navigate the Energy Landscape[C]//IEEE 2016 Fourth InternationalConference on 3D Vision (3DV).Stanford,CA,USA,2016:323-332.
[15] SHOTTON J,GLOCKER B,ZACH C,et al.Scene Coordinate Regression Forests for Camera Relocalization in RGB-D Images[C]//IEEE Conference on Computer Vision and Pattern Recognition.IEEE Computer Society,2013:2930-2937.
[16] XIAO J,OWENS A,TORRALBA A.SUN3D:A Database of Big Spaces Reconstructed Using SfM and Object Labels[C]//IEEE International Conference on Computer Vision.IEEE Computer Society,2013:1625-1632.
[17] The Stanford 3D Scanning Repository[EB/OL].(2014-08-19).www.graphics.st-anford.edu.