Efficient Mesh Reconstruction and Texturing of Oracle Bones.

Area of Science:

• Archaeological Sciences
• Computer Vision
• Digital Heritage

Background:

• High-fidelity 3D digitization of small, detailed cultural heritage objects is challenging.
• Existing methods like Structure-from-Motion (SfM), Light Detection and Ranging (LiDAR), and RGB-Depth have limitations in geometric density, data overhead, and color mapping.
• Integrating multi-shading techniques for surface texture extraction into cohesive meshes remains difficult.

Purpose of the Study:

• To develop a robust 3D digitization framework for archaeological artifacts.
• To address limitations in existing reconstruction workflows for capturing fine details.
• To improve geometric accuracy and texture consistency in digital models of cultural heritage objects.

Main Methods:

• A hybrid camera pose estimation strategy combining semi-automatic alignment with Iterative Closest Point (ICP)-based refinement.
• An efficient mesh-based representation utilizing vertex-level coloring for detailed geometry and consistent texturing.
• An integrated hardware-assisted workflow for controlled acquisition of small archaeological artifacts.

Main Results:

• Achieved a mean geometric reconstruction error of approximately 0.075 mm on Oracle Bone artifacts.
• Demonstrated a Hausdorff distance of 1 mm, indicating high fidelity in reconstruction.
• Successfully preserved fine inscription geometry and provided consistent texturing with compact storage.

Conclusions:

• The proposed digitization framework effectively reconstructs small, detailed archaeological artifacts with high geometric accuracy.
• The hybrid pose estimation and mesh representation offer robust alignment and efficient data handling.
• The integrated workflow is tailored for digitizing cultural heritage objects, particularly Oracle Bones, under controlled conditions.