Optimizing Fingernail Imaging Calibration for 3D Force Magnitude Prediction.

Area of Science:

• Biomedical Engineering
• Biomechanics
• Computer Vision

Background:

• Accurate fingerpad force prediction is crucial for human-computer interaction and robotics.
• Fingernail imaging offers a non-invasive method for capturing biomechanical data.
• Optimization of imaging parameters is essential for reliable force prediction.

Purpose of the Study:

• To optimize a fingernail imaging system for predicting fingerpad force.
• To evaluate the impact of lighting, calibration grids, and prediction models on system performance.
• To identify optimal parameters for system calibration.

Main Methods:

• Investigated the effects of white and green LED lighting on registration and force prediction.
• Compared Cartesian and cylindrical calibration grids for their impact on accuracy.
• Evaluated five different force prediction models, including a principal component regression approach.

Main Results:

• LED lighting color (white vs. green) showed no statistically significant difference in performance.
• Cartesian and cylindrical calibration grids yielded similar registration and force prediction results.
• The EigenNail Magnitude Model, utilizing pixel intensity eigenvectors, demonstrated the highest force prediction accuracy.

Conclusions:

• Optimal parameter choices for fingernail imaging system calibration have been identified.
• The EigenNail Magnitude Model provides accurate, multi-directional force estimation (RMS error 0.55 ± 0.02 N).
• This optimized system enhances the potential for precise fingerpad force prediction.