数据与计算发展前沿 ›› 2021, Vol. 3 ›› Issue (4): 30-43.doi: 10.11871/jfdc.issn.2096-742X.2021.04.003

• 可视化与可视分析专题 • 上一篇    下一篇

基于虚拟视点的大规模数值模拟的可交互式原位可视化

石晓霖1(),毕重科1,*(),黄元琪1(),邓亮2(),王岳青2(),王昉2()   

  1. 1.天津大学智能与计算学部,天津 300350
    2.中国空气动力研究与发展中心,四川 绵阳 621000
  • 收稿日期:2021-06-10 出版日期:2021-08-20 发布日期:2021-08-30
  • 通讯作者: 毕重科
  • 作者简介:石晓霖,天津大学智能与计算学部,硕士研究生,主要研究方向为科学可视化。
    本文主要承担算法实现和论文修改工作。
    SHI Xiaolin is currently a master's student in the College of Intelligence and Computing at Tianjin University, China. His research field includes scientific visualization.
    In this paper, he is mainly responsible for algorithms implemen-tation and paper revision.
    E-mail: sxl@tju.edu.cn|毕重科,天津大学智能与计算学部,副教授,主要研究科学可视化和高性能计算。
    本文主要承担论文总体框架,并指导基于特征区域的三维重建方法的实现。
    BI Chongke is currently an associate professor in the College of Intelligence and Computing at Tian-jin University, China. His research interests include scientific visualization and high performance computing.
    In this paper, he is responsible for the design for the framework of the paper; and he is the director for the design and implemen-tation of the 3D reconstruction method.
    E-mail: bichongke@tju.edu.cn|黄元琪,天津大学智能与计算学部,硕士,主要研究方向为科学可视化和高性能计算。
    本文主要承担文献调研和实验。
    HUANG Yuanqi received his master's degree from the College of Intelligence and Computing at Tianjin University, China. His research inter-ests include scientific visualization and high performance com-puting.
    In this paper, he is mainly responsible for the survey and experimental evaluation.
    E-mail: 742023742@qq.com|邓亮,中国空气动力研究与发展中心计算空气动力研究所,助理研究员,主要研究方向为高性能计算和计算流体力学。
    本文主要承担指导原位可视化框架的实现。
    DENG Liang is currently an assistant researcher of Computational Aerodynamics Institute, China Aerodynamics Research and Development Center. His research interests include high performance computing and CFD.
    In this paper, he is the director for design and implementation of the in-situ visualization framework.
    E-mail: dengliang11@nudt.edu.cn|王岳青,中国空气动力研究与发展中心计算空气动力研究所,助理研究员,主要研究方向为高性能计算和计算流体力学。
    本文主要承担指导虚拟视点合成方法的实现。
    WANG Yueqing is currently an assistant researcher of Comput-ational Aerodynamics Institute, China Aerodynamics Research and Development Center. His research interests include high performance computing and CFD.
    In this paper, he is the director for design and implementation of the virtual viewpoint synthesis method.
    E-mail: yqwang2013@163.com|王昉,中国空气动力研究与发展中心计算空气动力研究所,高级工程师,主要研究方向为高性能计算和计算流体力学。
    本文主要承担指导基于特征点的图像拼接方法的实现。
    WANG Fang is currently a senior engineer of Computational Aerodynamics Institute, China Aerodyna-mics Research and Development Center. His research interests include high performance computing and CFD.
    In this paper, he is the director for design and implementation of the image stitching method.
    E-mail: wwangfang@sina.com
  • 基金资助:
    国家数值风洞工程项目;国家重点研发计划(2021YFE0108400)

Virtual Viewpoint Based Interactive In-situ Visualization of Large-Scale Simulation

SHI Xiaolin1(),BI Chongke1,*(),HUANG Yuanqi1(),DENG Liang2(),WANG Yueqing2(),WANG Fang2()   

  1. 1. College of Intelligence and Computing, Tianjin University, Tianjin 300350, China
    2. China Aerodynamics Research and Development Center, Mianyang, Sichuan 621000, China
  • Received:2021-06-10 Online:2021-08-20 Published:2021-08-30
  • Contact: BI Chongke

摘要:

【目的】原位可视化是解决大规模数值模拟结果数据难以输出到磁盘进行后处理可视化的有效途径,但是当前的原位可视化主要是为专家提供浏览模式,难以进行交互式探索。所以研究原位可视化的交互技术对帮助专家分析数值模拟结果具有重要意义。【方法】本文提出了一种基于虚拟视点的可交互式原位可视化方法。为了解决原位可视化图像采样不全的问题,我们设计了基于深度图的虚拟视点合成方法,补充了用于三维重建的虚拟视点图像。然后为了展示数值模拟结果的整体特征,将生成的虚拟视点图像和原始图像进行图像拼接。最后对拼接而成的完整图像进行三维重建生成了可供交互的模型,解决了原位可视化图像难以帮助专家在空间上交互的问题。【结果】我们使用了二维和三维等不同数值模拟生成的原位可视化图像进行实验验证,证明了本文提出的可交互式原位可视化方法的有效性。【局限】图像处理技术在不断发展,未来希望对原位可视化数据与图像处理的交互式结合进行更多的探索。【结论】本文设计了一种基于虚拟视点的可交互式原位可视化方法,可以有效地帮助专家准实时观察和分析大规模数值模拟所有中间结果的细节。

关键词: 原位可视化, 交互式操作, 图像处理

Abstract:

[Objective] In-situ visualization is an effective way to solve the problem of outputting large-scale numerical simulation results to the hard disk for post-processing visualization. However, the current in-situ visualization methods mainly provide browsing mode for experts, which fails to conduct interactive exploration. Therefore, the study of interactive technology for in-situ visualization is essential for helping experts in analyzing the results of numerical simulations. [Methods] We present an interactive in-situ visualization method based on virtual viewpoint to help experts visually explore numerical simulations. Firstly, in order to resolve the problem of incomplete sampling of in-situ visualization images, a virtual viewpoint synthesis method is designed to supplement virtual viewpoint images for 3D reconstruction. Then, based on raw images and the generated virtual viewpoint images, an image stitching method is proposed to show the overall characteristics of in-situ visualization images. Finally, the interactive model can be generated through 3D reconstruction from the stitched images. [Results] In order to demonstrate the effectiveness of the proposed interactive in-situ model, several different numerical simulations are carried out. As a result, the generated in-situ visualization results can be interactively explored by domain scientists. [Limitations] Because that the image processing technology is constantly evolving, we hope to conduct more explorations on combination of in-situ visualization image data and interactive image processing in the future. [Conclusions] In this paper, a virtual viewpoint-based interactive in-situ visualization method is proposed. It can successfully help experts to interactively explore the intermediate output data of large-scale simulation.

Key words: in-situ visualization, interactive operation, image processing