Multiple regression based imputation for individualizing template human model from a small number of measured...

Area of Science:

• Anthropometry
• 3D Human Modeling
• Biomechanical Engineering

Background:

• Creating accurate individual human models typically requires extensive 3D scanning or numerous measurements.
• Existing methods can be time-consuming and may not capture all necessary dimensions for detailed modeling.

Purpose of the Study:

• To develop and validate a novel method for estimating a complete set of human body dimensions from a limited subset of measurements.
• To reduce the complexity and time required for individual human model creation.

Main Methods:

• Utilized a human dimension database to solve multiple regression equations.
• Trained regression models with subset dimensions as explanatory variables and full set dimensions as the objective variable.
• Employed leave-one-out cross-validation to assess the accuracy of estimated dimensions.

Main Results:

• The regression-based method accurately estimated full body dimensions from a small subset.
• Average Mean Absolute Errors (MAE) for non-measured dimensions were 4.58% (hand), 4.42% (foot), and 3.54% (whole body).
• MAE for measured dimensions was 0.00%, confirming high precision for available data.

Conclusions:

• The proposed method offers an efficient and accurate approach for individual human model personalization.
• This technique significantly reduces the number of required measurements for creating detailed human models.
• The findings have implications for fields requiring precise anthropometric data, such as ergonomics and virtual reality.