帮助 关于我们

返回检索结果

基于平面补丁的自适应八叉树三维图像重建
Adaptive octree 3D image reconstruction based on plane patch

查看参考文献28篇

姚程 1,2 *   马彩文 1,2  
文摘 提出了一种基于平面补丁的自适应八叉树卷积神经网络(Octree Convolutional Neural Networks,O-CNN),用于进行有效的三维形状编码和解码。不同于基于体素或基于八叉树的卷积神经网络(Convolutional Neural Networks, CNN),以相同的分辨率表示具有体素的三维形状,O-CNN可自适应地表示具有不同层次的八叉树节点的三维形状,并使用平面补丁对每个八叉树节点内的三维形状进行建模。依据这种自适应表示设计了一种用于编码和解码三维形状的自适应O-CNN编码器和解码器。自适应O-CNN编码器将平面补丁法线和位移作为输入,仅在每个级别的八叉树节点上执行三维卷积操作,而自适应O-CNN解码器则推断每个层次的八叉树节点的形状占有率和细分状态,并估计每个最佳叶八叉树节点的平面法线和位移。通过对单个图像的形状预测验证了自适应O-CNN的生成任务的效率和有效性,倒角距离误差为0.274,低于OctGen的倒角距离误差0.294,取得了更好的重建效果。作为3D形状分析和生成的通用框架,基于平面补丁的自适应O-CNN不仅减少了内存和计算成本,而且比现有的3D-CNN方法具有更好的形状生成能力。
其他语种文摘 In this study,an adaptive octree convolutional neural network based on plane patches is proposed for effective 3D shape encoding and decoding. Unlike volume-based or octree-based convolutional neural network(CNN)methods,which represent 3D shapes with the same voxel resolution,the proposed method can use planes and adaptively represent the 3D shapes of octree nodes with different levels. The patch models the 3D shape within each octree node,whereby the patch-based adaptive representation is utilized in the proposed adaptive patch octree convolutional neural network(O-CNN)encoder and decoder for the encoding and decoding of 3D shapes. The adaptive patch O-CNN encoder takes the plane patch normal and displacement as input and performs three-dimensional convolution on the octree nodes of each level,whereas the adaptive patch O-CNN decoder infers each level. The shape occupancy rate and subdivision state of the octree node as well as the best plane normal and displacement of each leaf octree node are estimated. As a general framework for 3D shape analysis and generation,adaptive patch O-CNN not only reduces memory and computational costs but also exhibits better shape generation capabilities than existing 3D-CNN methods. Shape prediction is performed using a single image to verify the efficiency and effectiveness of the generation task of the adaptive O-CNN. The chamfer distance error is 0.274,which is lower than that of OctGen(0.294),resulting in a better reconstruction effect.
来源 光学精密工程 ,2022,30(9):1113-1122 【核心库】
DOI 10.37188/OPE.20223009.1113
关键词 计算机视觉 ; 三维重建 ; 卷积神经网络 ; 神经网络
地址

1. 中国科学院西安光学精密机械研究所, 陕西, 西安, 710119  

2. 中国科学院大学, 北京, 100049

语种 中文
文献类型 研究性论文
ISSN 1004-924X
学科 机械、仪表工业;自动化技术、计算机技术
基金 吉林省自然科学基金
文献收藏号 CSCD:7301339

参考文献 共 28 共2页

1.  范丽丽. 基于深度卷积神经网络的目标检测研究综述. 光学精密工程,2020,28(5):1152-1164 被引 54 次    
2.  Lun Z L. 3D shape reconstruction from sketches via multi-view convolutional networks. 2017 International Conference on 3D Vision(3DV),2017:67-77 被引 1 次    
3.  潘仙张. 多模深度卷积神经网络应用于视频表情识别. 光学精密工程,2019,27(4):963-970 被引 13 次    
4.  Niu Z J. DeepLabbased spatial feature extraction for hyperspectral im age classification. IEEE Geoscience and Remote Sensing Letters,2019,16(2):251-255 被引 6 次    
5.  Wu J J. Learning a probabilistic latent space of object shapes via 3D generative-adversarial modeling. NIPS'16:Proceedings of the 30th International Conference on Neural Information Processing Systems,2016:82-90 被引 1 次    
6.  Wu Z R. 3D ShapeNets: a deep representation for volumetric shapes. 2015 IEEE Conference on Computer Vision and Pattern Recognition,2015:1912-1920 被引 16 次    
7.  Bronstein M M. Geometric deep learning:going beyond euclidean data. IEEE Signal Processing Magazine,2017,34(4):18-42 被引 66 次    
8.  Groueix T. AtlasNet:a papier-mche approach to learning 3D surface generation. arXiv: 1802.05384 [cs. CV],2018 被引 1 次    
9.  Kato H. Neural 3D mesh renderer. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition,2018:3907-3916 被引 1 次    
10.  吴笑天. 基于区域选择网络的图像复原及其在计算成像中的应用. 光学精密工程,2021,29(4):864-876 被引 4 次    
11.  Attene M. Polygon mesh repairing. ACM Computing Surveys,2013,45(2):1-33 被引 15 次    
12.  Boscaini D. Learning class-specific descriptors for deformable shapes using localized spectral convolutional networks. Computer Graphics Forum,2015,34(5):13-23 被引 9 次    
13.  Brock A. Generative and discriminative voxel modeling with convolutional neural networks. arXiv: 1608.04236 [cs. CV],2016 被引 1 次    
14.  Choy C B. 3DR2N2: a unified approach for single and multi-view 3D object reconstruction. Computer Vision-ECCV 2016,2016:628-644 被引 4 次    
15.  Graham B. Sparse 3D convolutional neural networks. Proceedings ofthe British Machine Vision Conference 2015,2015:150.1-150.9 被引 1 次    
16.  Hane C. Hierarchical surface prediction. IEEE Transactions on Pattern Analysis and Machine Intelligence,2020,42(6):1348-1361 被引 2 次    
17.  Fuhrmann S. Floating scale surface reconstruction. ACM Transactions on Graphics,2014,33(4):1-11 被引 8 次    
18.  Deng J. ImageNet: a large-scale hierarchical image database. 2009 IEEE Conference on Computer Vision and Pattern Recognition,2009:248-255 被引 226 次    
19.  Charles R Q. Point-Net:deep learning on point sets for 3D classification and segmentation. 2017 IEEE Conference on Computer Vision and Pattern Recognition,2017:77-85 被引 26 次    
20.  Achlioptas P. Learning representations and generative models for 3D point clouds. ICML,2017 被引 1 次    
引证文献 1 篇

1 张新荣 基于转动式二维激光扫描仪和多传感器的三维重建方法 中国光学（中英文）,2023,16(3):663-672
被引 1 次

显示所有1篇文献

论文科学数据集
PlumX Metrics
相关文献

 作者相关
 关键词相关
 参考文献相关

版权所有 ©2008 中国科学院文献情报中心 制作维护：中国科学院文献情报中心
地址：北京中关村北四环西路33号 邮政编码：100190 联系电话：(010)82627496 E-mail:cscd@mail.las.ac.cn 京ICP备05002861号-4 | 京公网安备11010802043238号