CAREER: A Parallel and Efficient Computational Framework for Unified Volumetric Meshing in Large-Scale 3D/4D Anisotropy

职业生涯：大规模 3D/4D 各向异性中统一体积网格划分的并行高效计算框架

• 批准号: 1845962
• 负责人: Zichun Zhong
• 金额: $ 50万
• 依托单位: Wayne State University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-03-15 至 2025-02-28
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1845962&HistoricalAwards=false
• 关键词: CAREER; Parallel; Efficient; Computational; Framework

This proposal develops a computational framework that helps the domain scientists who employ advanced cyberinfrastructure ecosystem (e.g., for engineering, manufacturing, healthcare, etc.) to realistically and efficiently reconstruct, visualize, and analyze 3D and 4D (space-time) volumetric objects with complex geometric structures and highly anisotropic properties (such properties are characterized by the presence of specified orientations and aspect ratios in the system). For example, in mechanical engineering, it is necessary to interactively design and model mechanical parts with user-required high-quality measures and standards. The computational framework enables fabrication of such mechanical parts with specified microstructure that can be efficiently produced to sustain much stronger stress and strain compared with those without endowing such properties, which leads to significant impact on the next-generation mechanical component design. As an integral part of the PI's career development, the educational plan emphasizes on the integration of education and research in different aspects through the PI's new "3D hands-on" education philosophy for K-12, undergraduate and graduate students. This project thus serves the national interest, as stated by NSF's mission: to promote the progress of science; to advance the national health, prosperity and welfare. The research goal of this project focuses on a computational framework for anisotropic volumetric meshing, a foundational as well as translational research impacting a broad range of scientific domains. The capability and usability of the meshing framework are evaluated by investigating fabrication of objects with internal microstructures and construction of anisotropic volumetric models to capture the organ and tissue shape. This work has the following primary components: (1) Computing high-dimensional geometric embedding based on Nash theorem in parallel: the computational realization of high-dimensional geometric embedding makes modeling complex objects with multiple tensor features being built and solved in parallel in a large linear system. (2) Modeling multi-shape of mesh element in a unified particle framework: the particle system flexibly and effectively generates high-quality honeycomb, tetrahedral, and hexahedral (grid) patterns, which are exactly designed for meshing structure. The optimization procedure is easily formulated for parallelism in the high-dimensional space. (3) Generating 3D/4D anisotropic mesh in parallel: the final multi-shape anisotropic meshes are computed in parallel in the high-dimensional space with simple Euclidean computations under the isotropic metric. The primary outcome of this project is a 3D/4D-ParaAnisoMesh system.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

该提案开发了一个计算框架，帮助采用先进网络基础设施生态系统（例如，用于工程、制造、医疗保健等）的领域科学家通过以下方法真实有效地重建、可视化和分析 3D 和 4D（时空）体积对象：复杂的几何结构和高度各向异性的特性（这些特性的特征是系统中存在指定的方向和纵横比）。例如，在机械工程中，需要按照用户要求的高质量措施和标准对机械零件进行交互式设计和建模。该计算框架能够制造具有特定微观结构的机械零件，与不具有这种特性的机械零件相比，这些机械零件可以有效地生产以承受更强的应力和应变，这对下一代机械零件设计产生重大影响。作为PI职业发展的重要组成部分，该教育计划强调通过PI针对K-12、本科生和研究生的全新“3D动手”教育理念，强调教育与研究在不同方面的融合。因此，该项目符合国家利益，正如 NSF 的使命所言：促进科学进步；促进国民健康、繁荣和福利。该项目的研究目标侧重于各向异性体积网格划分的计算框架，这是影响广泛科学领域的基础研究和转化研究。 通过研究具有内部微观结构的物体的制造以及捕获器官和组织形状的各向异性体积模型的构建来评估网格框架的功能和可用性。这项工作有以下主要组成部分：（1）基于纳什定理的并行计算高维几何嵌入：高维几何嵌入的计算实现使得对具有多个张量特征的复杂对象进行建模并在大范围内并行构建和求解线性系统。 （2）在统一的粒子框架中对多种形状的网格单元进行建模：粒子系统灵活有效地生成高质量的蜂窝状、四面体和六面体（网格）图案，这些图案正是为网格结构而设计的。优化过程很容易制定用于高维空间中的并行性。 (3)并行生成3D/4D各向异性网格：在各向同性度量下通过简单的欧几里德计算在高维空间中并行计算最终的多形状各向异性网格。该项目的主要成果是 3D/4D-ParaAnisoMesh 系统。该奖项反映了 NSF 的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

DeepOrganNet: On-the-Fly Reconstruction and Visualization of 3D / 4D Lung Models from Single-View Projections by Deep Deformation Network

DeepOrganNet：通过深度变形网络从单视图投影实时重建和可视化 3D / 4D 肺模型

• 发表时间: 2020-01
• 期刊: IEEE transactions on visualization and computer graphics
• 影响因子: 5.2
• 作者: Wang, Yifan;Zhong, Zichun;Hua, Jing
• 通讯作者: Hua, Jing

VC-Net: Deep Volume-Composition Networks for Segmentation and Visualization of Highly Sparse and Noisy Image Data

VC-Net：用于高度稀疏和噪声图像数据的分割和可视化的深度体合成网络

• DOI: 10.1109/tvcg.2020.3030374
• 发表时间: 2021-01
• 期刊: IEEE transactions on visualization and computer graphics
• 影响因子: 5.2
• 作者: Wang, Yifan;Yan, Guoli;Zhu, Haikuan;Buch, Sagar;Wang, Ying;Haacke, Ewart Mark;Hua, Jing;Zhong, Zichun
• 通讯作者: Zhong, Zichun

DiffSVR: Differentiable Neural Implicit Surface Rendering for Single-View Reconstruction with Highly Sparse Depth Prior

DiffSVR：用于具有高度稀疏深度先验的单视图重建的可微神经隐式表面渲染

• DOI: 10.1016/j.cad.2023.103604
• 发表时间: 2023-08-01
• 期刊: Comput. Aided Des.
• 作者: Artem Komarichev;Jing Hua;Z. Zhong
• 通讯作者: Z. Zhong

Multi-Derivative Physical and Geometric Convolutional Embedding Networks for Skeleton-based Action Recognition

用于基于骨架的动作识别的多导数物理和几何卷积嵌入网络

• 发表时间: 2021-01
• 期刊: Computer aided geometric design
• 影响因子: 1.5
• 作者: Yan, Guoli;Hua, Michelle;Zhong, Zichun
• 通讯作者: Zhong, Zichun

JointFontGAN: Joint Geometry-Content GAN for Font Generation via Few-Shot Learning

JointFontGAN：通过少样本学习生成字体的联合几何内容 GAN

• 发表时间: 2020-01
• 期刊: ACM Multimedia (ACM MM
• 作者: Xi, Yankun;Yan, Guoli;Hua, Jing;Zhong, Zichun
• 通讯作者: Zhong, Zichun