基于隐式描述符的三维模型对应关系计算

doi:10.19678/j.issn.1000-3428.0061650

摘要/Abstract

摘要： 三维模型对应关系计算在自动驾驶、虚拟现实、智能交通等领域得到广泛关注与应用。三维模型在几何结构和尺度发生很大变化时，低层次几何信息描述符所提取的特征将不足，从而使得对应关系计算结果准确率不高。为此，提出一种通过引入先验知识来完成三维模型对应关系计算的方法。利用深度学习网络模仿人类计算先验知识，以对模型各部分之间的几何相似性进行编码，解决模型在各部分发生显著变化时无法应用低层次几何信息计算模型间对应关系的问题。使用多视图卷积神经网络对模型各部分相应的视图进行预分割并标记，根据模型对应表面点之间的相似度隐式地计算数据驱动描述符，在数据驱动描述符的指导下计算最终的三维模型对应关系。实验结果表明，相较基于先验知识的计算方法，该方法能提高三维模型对应关系计算结果的准确率，且可有效降低测地误差。

关键词: 三维模型, 对应关系, 多视图卷积神经网络, 数据驱动描述符, 先验知识

Abstract: The calculation of correspondences between 3D models has been widely studied and applied in fields such as autonomous vehicles, virtual reality, and intelligent transportation.However, if the geometric structure and scale of the 3D models compared differ substantially, features extracted by low-level geometric information descriptors are insufficient, as is the accuracy of the results of such correspondence calculations.Therefore, this paper proposes a method to calculate the correspondence of 3D models by introducing a priori knowledge.A deep learning network is used to simulate human prior knowledge to encode geometric similarities between the parts of two models.The proposed method solves the problem that low-level geometric information cannot be used to calculate correspondence relationships between models when their parts differ significantly.A multi-view convolutional neural network is used to pre-segment and mark the corresponding views of each part of the model, to implicitly calculate a data-driven descriptor according to the similarity between corresponding surface points, and finally to calculate the output correspondence between the two 3D models under the guidance of the data-driven descriptor.The experimental results show that the proposed method was able to improve the accuracy of the results of calculations of correspondence relationships between 3D models compared with a calculation method based on a priori knowledge, and it can effectively reduce the geodetic error.

Key words: 3D model, correspondence, multi-view convolutional neural network, data-driven descriptor, prior knowledge

中图分类号:

TP391

HAYTHEM Alhag, 杨军. 基于隐式描述符的三维模型对应关系计算[J]. 计算机工程, 2022, 48(5): 229-234.

HAYTHEM Alhag, YANG Jun. Calculation of 3D Model Correspondence Based on Implicit Descriptor[J]. Computer Engineering, 2022, 48(5): 229-234.

https://www.ecice06.com/CN/Y2022/V48/I5/229

图/表 5

20220806174437

20220806174440

20220806174444

20220806174447

20220806174450

参考文献

[1] VAN KAICK O, ZHANG H, HAMARNEH G, et al.A survey on shape correspondence[J].Computer Graphics Forum, 2011, 30(6):1681-1707.
[2] VAN KAICK O, TAGLIASACCHI A, SIDI O, et al.Prior knowledge for part correspondence[J].Computer Graphics Forum, 2011, 30(2):553-562.
[3] TANGELDER J W H, VELTKAMP R C.A survey of content based 3D shape retrieval methods[J].Multimedia Tools and Applications, 2007, 39(3):441-471.
[4] IYER N, JAYANTI S, LOU K Y, et al.Three-dimensional shape searching:state-of-the-art review and future trends[J].Computer-Aided Design, 2005, 37(5):509-530.
[5] ATTENE M, BIASOTTI S, MORTARA M, et al.Computational methods for understanding 3D shapes[J].Computers & Graphics, 2006, 30(3):323-333.
[6] ZHANG H, SHEFFER A, COHEN-OR D, et al.Deformation-driven shape correspondence[C]//Proceedings of Symposium on Geometry Processing.Berlin, Germany:Springer, 2008:1431-1439.
[7] KIN-CHUNG AU O, TAI C L, COHEN-OR D, et al.Electors voting for fast automatic shape correspondence[J].Computer Graphics Forum, 2010, 29(2):645-654.
[8] HOFFMAN D D, RICHARDS W A.Parts of recognition[J].Cognition, 1984, 18(1/2/3):65-96.
[9] KREAVOY V, JULIUS D, SHEFFER A.Model composition from interchangeable components[C]//Proceedings of the 15th Pacific Conference on Computer Graphics and Applications.Berlin, Germany:Springer, 2007:129-138.
[10] KATZ S, TAL A.Hierarchical mesh decomposition using fuzzy clustering and cuts[J].ACM Transactions on Graphics, 2003, 22(3):954-961.
[11] LIU R, ZHANG H.Mesh segmentation via spectral embedding and contour analysis[J].Computer Graphics Forum, 2007, 26(3):385-394.
[12] SHAPIRA L, SHAMIR A, COHEN-OR D.Consistent mesh partitioning and skeletonisation using the shape diameter function[J].The Visual Computer, 2008, 24(4):249-259.
[13] GOLOVINSKIY A, FUNKHOUSER T.Consistent segmentation of 3D models[J].Computers & Graphics, 2009, 33(3):262-269.
[14] XU K, LI H H, ZHANG H, et al.Style-content separation by anisotropic part scales[J].ACM Transactions on Graphics, 2010, 29(6):184.
[15] SIMARI P, NOWROUZEZAHRAI D, KALOGERAKIS E, et al.Multi-objective shape segmentation and labeling[EB/OL].[2021-04-05].http://diglib2.eg.org/EG/DL/CGF/volume28/issue5/v28i5pp1415-1425.pdf.abstract.pdf.
[16] KALOGERAKIS E, HERTZMANN A, SINGH K.Learning 3D mesh segmentation and labeling[C]//Proceedings of SIGGRAPH'10.New York, USA:ACM Press, 2010:1-12.
[17] XIANG R, LAI R J, ZHAO H K.Efficient and robust shape correspondence via sparsity-enforced quadratic assignment[EB/OL].[2021-04-05].https://arxiv.org/abs/2003.08680.
[18] HALIMI O, LITANY O, RODOLÀ E R, et al.Unsupervised learning of dense shape correspondence[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2019:4365-4374.
[19] GAO Z, WANG D Y, XUE Y B, et al.3D object recognition based on pairwise multi-view convolutional neural networks[J].Journal of Visual Communication and Image Representation, 2018, 56:305-315.
[20] SU H, MAJI S, KALOGERAKIS E, et al.Multi-view convolutional neural networks for 3D shape recognition[C]//Proceedings of IEEE International Conference on Computer Vision.Washington D.C., USA:IEEE Press, 2015:945-953.
[21] HUANG H B, KALOGERAKIS E, CHAUDHURI S, et al.Learning local shape descriptors from part correspondences with multiview convolutional networks[J].ACM Transactions on Graphics, 2018, 37(1):6.
[22] YI L, KIM V G, CEYLAN D, et al.A scalable active framework for region annotation in 3D shape collections[J].ACM Transactions on Graphics, 2016, 35(6):210.
[23] KRIZHEVSKY A, SUTSKEVER I, HINTON G E.ImageNet classification with deep convolutional neural networks[J].Communications of the ACM, 2017, 60(6):84-90.
[24] HADSELL R, CHOPRA S, LECUN Y.Dimensionality reduction by learning an invariant mapping[C]//Proceedings of CVPR'06.Washington D.C., USA:IEEE Press, 2006:1735-1742.
[25] RUSSAKOVSKY O, DENG J, SU H, et al.ImageNet large scale visual recognition challenge[J].International Journal of Computer Vision, 2015, 115(3):211-252.
[26] EZUZ D, HEEREN B, AZENCOT O, et al.Elastic correspondence between triangle meshes[J].Computer Graphics Forum, 2019, 38(2):121-134.
[27] CHEN X B, GOLOVINSKIY A, FUNKHOUSER T.A benchmark for 3D mesh segmentation[J].ACM Transactions on Graphics, 2009, 28(3):73.

选择文件类型/文献管理软件名称

选择包含的内容