期刊
KNOWLEDGE-BASED SYSTEMS
卷 242, 期 -, 页码 -出版社
ELSEVIER
DOI: 10.1016/j.knosys.2022.108395
关键词
3D reconstruction; 3D object reconstruction; Single-view reconstruction; Point cloud reconstruction; Structured point cloud reconstruction
资金
- National Natural Science Foundation of China [61902159, 62106227, 61771146]
- Dual Creative Doctors of Jiangsu Province of China
This paper proposes a Part-Wise AtlasNet method based on the architecture of AtlasNet, which imposes constraints on the local structures of 3D objects by restricting each neural network to reconstructing a specific part of the object. The experimental results demonstrate that the proposed method generates structured point clouds with higher visual quality and better performance in 3D point cloud generation from a single image compared to other methods.
Learning to generate three dimensional (3D) point clouds from a single image remains a challenging task. Numerous approaches with encoder-decoder architectures have been proposed. However, these methods are hard to realize structured reconstructions and usually lack constraints on the local structures of 3D objects. AtlasNet as a representative model of 3D reconstruction consists of many branches, and each branch is a neural network used to reconstruct one local patch of a 3D object. However, the neural networks in AtlasNet and the patches of 3D objects are not in one-to-one correspondence before training. This case is not conducive to adding some reconstruction constraints to the local structures of 3D objects. Based on the architecture of AtlasNet, we propose Part-Wise AtlasNet in which each neural network is only responsible for reconstructing one specific part of a 3D object. This kind of restriction facilitates imposition of several local constraints on the final reconstruction loss, hence better recovering 3D objects with finer local structures. Both the qualitative results and quantitative analysis show that the variants of the proposed method with the local reconstruction losses generate structured point clouds with a higher visual quality and achieve better performance than other methods in 3D point cloud generation from a single image. (C)& nbsp;2022 Elsevier B.V. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据