基于卷积神经网络的点云配准方法

舒程珣; 何云涛; 孙庆科

doi:doi:10.3788/lop54.031001

激光与光电子学进展, 2017, 54 (3): 031001, 网络出版: 2017-03-08

基于卷积神经网络的点云配准方法下载： 2240次

Point Cloud Registration Based on Convolutional Neural Network

论文大纲

舒程珣 ^1,*何云涛 ¹孙庆科 ²

作者单位

¹ 北京航空航天大学电子信息工程学院, 北京 100191

² 河北远东通信系统工程有限公司, 河北石家庄 050200

图像处理点云配准深度学习卷积神经网络深度图像 image processing point cloud registration deep learning convolutional neural network depth image

AI 词云图 AI一句话精读 AI短摘要

注：本部分内容由 AI 自动生成，请您知悉。

摘要

点云配准是三维点云信息处理中的重要问题。传统点云配准方法计算量大，不利于实时计算与移动计算。针对传统点云配准方法存在的问题，提出了一种利用卷积神经网络进行点云配准的方法。首先计算点云的深度图像，利用卷积神经网络提取深度图像对的特征差，将深度图像对的特征差作为全连接网络的输入并计算点云配准参数，迭代地执行上述操作直至配准误差小于可接受阈值。实验结果表明，相比传统的点云配准方法，基于卷积神经网络的点云配准方法具有所需计算量小、配准效率高、对噪声点和异常点不敏感的优点。

Abstract

Point cloud registration is an important issue in 3D information processing. The traditional point cloud registration needs a huge amount of computation, thus it is not suitable for real-time and mobile computation. In order to solve the problem of traditional point cloud registration method, a method based on convolutional neural network is proposed. The depth image of point cloud is calculated and the differential feature vector of depth images extracted by the convolutional neural network is regarded as input of fully connected neural network to calculate registration parameters. Iteratively executing the above process until registration error is acceptable. Experimental results show that the point cloud registration based on convolutional neural network is simpler in computation, more efficient in registration rate, and less sensitive to noise and outlier than the traditional methods.

参考文献

[1] Nguyen T T,刘修国, 王红平,等. 基于激光扫描技术的三维模型重建［J］. 激光与光电子学进展, 2011, 48(8): 081201.

Nguyen T T,Liu Xiuguo, Wang Hongping, et al. 3D model reconstruction based on laser scanning technique［J］. Laser & Optoelectronics Progress, 2011, 48(8): 081201.

[2] Besl P J, McKay N D. Method for registration of 3D shapes［C］. Robotics-DL tentative, International Society for Optics and Photonics, 1992: 586-606.

[3] Chen Y, Medioni G. Object modelling by registration of multiple range images［J］. Image and Vision Computing, 1992, 10(3): 145-155.

[4] Blais G, Levine M D. Registering multiview range data to create 3D computer objects［C］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1995, 17(8): 820-824.

[5] 韦盛斌, 王少卿, 周常河, 等. 用于三维重建的点云单应性迭代最近点配准算法［J］. 光学学报, 2015, 35(5): 0515003.

Wei Shengbin, Wang Shaoqing, Zhou Changhe, et al. An iterative closest point algorithm based on biunique correspondence of point clouds for 3D reconstruction［J］. Acta Optica Sinica, 2015, 35(5): 0515003.

[6] 左超, 鲁敏, 谭志国, 等. 一种新的点云拼接算法［J］. 中国激光, 2012, 39(12): 1214004.

Zuo Chao, Lu Min, Tan Zhiguo, et al. A novel algorithm for registration of point clouds［J］. Chinese J Lasers, 2012, 39(12): 1214004.

[7] 伍梦琦, 李中伟, 钟凯, 等. 基于几何特征和图像特征的点云自适应拼接方法［J］. 光学学报, 2015, 35(2): 0215002.

Wu Mengqi, Li Zhongwei, Zhong Kai, et al. Adaptive point cloud registration method based on geometric features and photometric features［J］. Acta Optica Sinica, 2015, 35(2): 0215002.

[8] 黄源, 达飞鹏, 陶海跻. 一种基于特征提取的点云自动配准算法［J］. 中国激光, 2015, 42(3): 0308002.

Huang Yuan, Da Feipeng, Tao Haiqi. An automatic registration algorithm for point cloud based on feature extraction［J］. Chinese J Lasers, 2015, 42(3): 0308002.

[9] 张晓, 张爱武, 王致华. 基于改进正态分布变换算法的点云配准［J］. 激光与光电子学进展, 2014, 51(4): 041002.

Zhang Xiao, Zhang Aiwu, Wang Zhihua. Pointcloud registration based on improved normal distribution transform algorithm［J］. Laser & Optoelectronics Progress, 2014, 51(4): 041002.

[10] 余凯, 贾磊, 陈雨强, 等. 深度学习的昨天、今天和明天［J］. 计算机研究与发展, 2013, 50(9): 1799-1804.

Yu Kai, Jia Lei, Chen Yuqiang, et al. Deep learning: yesterday, today, and tomorrow［J］. Journal of Computer Research and Development, 2013, 50(9): 1799-1804

[11] Krizhevsky A, Sutskever I, Hinton G E. Imagenet classification with deep convolutional neural networks［C］. Advances in Neural Information Processing Systems, 2012, 25(2): 1097-1105.

[12] Deng J, Dong W, Socher R, et al. Image net: a large-scale hierarchical image database［C］. Proceedings of 2009 IEEE on Computer Vision and Pattern Recognition, 2009: 248-255.

[13] Mikolov T, Deoras A, Povey D, et al. Strategies for training large scale neural network language models［C］. IEEE Workshop on Automatic Speech Recognition and Understanding, 2011: 196-201.

[14] Silver D, Huang A, Maddison C J, et al. Mastering the game of Go with deep neural networks and tree search［J］. Nature, 2016, 529(7587): 484-489.

[15] Le Cun Y, Kavukcuoglu K, Farabet C. Convolutional networks and applications in vision［J］. ISCAS, 2010, 14(5): 253-256.

[16] Nair V, Hinton G E. Rectifiedlinear units improve restricted Boltzmann machines［C］. Proceedings of the 27th International Conference on Machine Learning, 2010: 807-814.

[17] Kaiser M, John M, Heimann T, et al. 2D/3D registration of TEE probe from two non-orthogonal C-arm directions［C］. International Conference on Medical Image Computing and Computer-Assisted Intervention, 2014: 283-290.

[18] Glorot X, Bengio Y. Understanding the difficulty of training deep feedforward neural networks［C］. International Conference on Artificial Intelligence and Statistics, 2010, 9: 249-256.

[19] Shilane P, Min P, Kazhdan M, et al. The princeton shape benchmark［C］. Proceedings of 2004 IEEE on Shape Modeling Application, 2004: 167-178.

[20] Jia Y, Shelhamer E, Donahue J, et al. Caffe: Convolutional architecture for fast feature embedding［C］. Proceedings of the 22nd ACM International Conference on Multimedia, 2014: 675-678.

[21] van de Kraats E B, Penney G P, Tomazevic D, et al. Standardized evaluation methodology for 2D-3D registration［J］. IEEE Transactions on Medical Imaging, 2005, 24(9): 1177-1189.

舒程珣, 何云涛, 孙庆科. 基于卷积神经网络的点云配准方法[J]. 激光与光电子学进展, 2017, 54(3): 031001. Shu Chengxun, He Yuntao, Sun Qingke. Point Cloud Registration Based on Convolutional Neural Network[J]. Laser & Optoelectronics Progress, 2017, 54(3): 031001.

基于卷积神经网络的点云配准方法下载： 2240次

关于本站 Cookie 的使用提示

全站搜索

基于卷积神经网络的点云配准方法 下载： 2240次

相关论文

相关资讯

关于本站 Cookie 的使用提示

全站搜索

基于卷积神经网络的点云配准方法下载： 2240次