Guidelines for Accurate Multi-Temporal Model Registration of 3D Scanned Objects.

Area of Science:

• Biomechanical Engineering
• 3D Imaging and Metrology
• Materials Science

Background:

• Quantifying object morphology changes over time requires accurate 3D models.
• 3D model registration aligns models using stable features, but feature selection guidelines are lacking.
• Accurate registration is vital for applications like wear analysis in biomechanics.

Purpose of the Study:

• To establish guidelines for selecting localized features for robust 3D model registration.
• To minimize registration error in multi-temporal 3D scanning.
• To ensure accuracy in quantifying morphological changes.

Main Methods:

• 3D optical scanning of a tibia bone replica at multiple time points.
• Alignment of 3D models using a cubic registration block with varied sizes and surfaces.
• Calculation of registration error based on surface variation between aligned models.

Main Results:

• Very low target registration errors were achieved when features occupied at least 4% of the scanning field of view.
• Selecting at least two orthogonal surfaces of the registration block minimized registration error.
• The study identified optimal parameters for accurate 3D model registration.

Conclusions:

• Selecting multiplanar features covering at least 4% of the scanning field of view is recommended for accurate 3D model registration.
• Utilizing at least two orthogonal surfaces enhances registration accuracy.
• These guidelines improve the reliability of 3D scanning for analyzing morphological changes.