A fast-modeling framework for personalized human body models based on a single image.

Area of Science:

• Biomechanics
• Computational modeling
• Human body modeling

Background:

• Finite element human body models (HBMs) are crucial for predicting injury responses in vehicle collisions.
• Personalized HBMs are essential for representing diverse populations but are difficult to create from single images.

Purpose of the Study:

• To develop a framework for personalizing HBMs using only a single image.
• To enable the creation of subject-specific HBMs for improved biomechanical simulations.

Main Methods:

• A framework was developed to estimate skin and skeletal point clouds, along with skeletal positions, from a single image.
• A point cloud registration-based mesh morphing technique was employed to personalize a baseline HBM.
• Eight personalized HBMs with varying biological characteristics were generated.

Main Results:

• The personalized HBMs exhibited comparable element quality to the baseline model.
• The mean geometric errors for the personalized finite element models (FEMs) were below 7 mm.
• Biomechanical response evaluations confirmed the acceptability of the generated personalized FEMs.

Conclusions:

• The proposed framework successfully generates personalized HBMs from a single image.
• This method allows for the creation of diverse, subject-specific HBMs for biomechanical research.
• The framework offers a viable solution for advancing personalized HBM development in automotive safety.