Scholarlink

We propose an interactive visualization and reconstruction system for gait pattern analysis that creates a more realistic visualization of the patient than contemporary stick figure-like representations. It supports thorough gait pattern analysis, and enables unbiased inter- and intra-patient comparisons as well as longitudinal studies. Our system takes as input 3D motion capture data, but can be further constrained using readily available metadata or layman-accessible measurements. A statistical shape model is then fitted to the motion capture and personal metadata. The result is an immediate and interactive visualization of an animated 3D twin. Our system handles different marker setups and can thus be applied to already existing data. We further show that we can infer realistic body models with only a few of markers. A survey with medical experts confirms the clinical applicability of our method.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

3D Body Twin: Improving Human Gait Visualizations Using Personalized Avatars

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Daniel Zieger,
  • Florian Güthlein,
  • Jann-Ole Henningson,
  • Verena Jakob,
  • Heiko Gaßner,
  • Julian Shanbhag,
  • Sophie Fleischmann,
  • Jörg Miehling,
  • Sandro Wartzack,
  • Björn Eskofier,
  • Jürgen Winkler,
  • Bernhard Egger,
  • Marc Stamminger

摘要

We propose an interactive visualization and reconstruction system for gait pattern analysis that creates a more realistic visualization of the patient than contemporary stick figure-like representations. It supports thorough gait pattern analysis, and enables unbiased inter- and intra-patient comparisons as well as longitudinal studies. Our system takes as input 3D motion capture data, but can be further constrained using readily available metadata or layman-accessible measurements. A statistical shape model is then fitted to the motion capture and personal metadata. The result is an immediate and interactive visualization of an animated 3D twin. Our system handles different marker setups and can thus be applied to already existing data. We further show that we can infer realistic body models with only a few of markers. A survey with medical experts confirms the clinical applicability of our method.